Master Art of Translation | The Satellite Conjunction, Space Debris & Orbital Data Translation System — How CCSDS CDMs, TCA, Covariance, Collision Probability and OMM Identifiers Keep Their Meaning

How to translate satellite conjunction warnings and orbital collision-risk data accurately

How do you translate a satellite conjunction data message (CDM), explain a space debris close-approach alert in another language, or preserve collision probability when a spacecraft warning moves between teams? Readers searching for satellite conjunction translation, orbital data translation, CCSDS CDM terminology, time of closest approach (TCA), covariance matrices, miss distance, and satellite collision-risk communication need a method that protects the actual evidence—not simply a fluent version of the words.

The hidden problem is that one small phrase can silently change the meaning of an operational record. A predicted close approach can become an asserted collision; a conditional probability can be mistaken for certainty; an orbital estimate can be presented as a measurement; a six-digit object identifier can disappear in a five-digit legacy feed. The result may look professional while changing the state, object, time, uncertainty or decision authority represented by the source.

This complete, world-facing guide explains how to translate and review conjunction warnings, space situational awareness summaries, collision-avoidance terminology, CCSDS Conjunction Data Messages, TLE and Orbit Mean-Elements Message (OMM) records, and their multilingual public explanations. It offers a repeatable preservation method, detailed technical mechanisms, worked fictional diagnostics, failure-mode checks, transfer exercises and source routes. It is an educational translation guide, not a collision-prediction service, flight-dynamics computation, maneuver instruction or source of mission-specific operating authority.

Architecture: EDKSG-TRANS-MASTER-WORLD-750. This new specialist child concerns the language, metadata and evidence integrity of satellite conjunction, tracking and orbital-risk records. It deliberately does not replace the general Master Art of Translation | The Complete System for Moving Meaning Between Languages or the existing space missions and satellite operations translation system, which retains the broader owner role for spacecraft telemetry, telecommands, mission planning and remote sensing.

Quick answer: what must a correct translation preserve?

Preserve the object identifiers; the issuing authority; creation and event timestamps; time scale; reference frame; units; version; whether data are measured, estimated or predicted; uncertainty and covariance meaning; miss-distance and probability definitions; the difference between screening, assessment and mitigation; the operational status of any maneuver; and every caveat about stale, missing or provisional input. A readable sentence that loses one of these can be scientifically or operationally wrong.

For a reader arriving with a particular warning, the fastest safe route is to identify the message type and version, locate the two objects and the time of closest approach, check the event state and its covariance, then write a plain-language explanation that states what the model predicts and what it does not establish. Keep raw numeric values and machine-readable field names intact. Translate the surrounding explanation, not a machine parser’s keys.

A field guide to this longform

• The hidden problem: assertions are not instructions
• What a conjunction message actually carries
• Technical translation mechanisms
• A repeatable translation and QA workflow
• Failure-mode decision table
• Worked fictional diagnostics
• Transfer workshops
• Terminology and quick definitions
• Primary references and next learning routes

The hidden problem: an orbital warning is an evidence object, not a dramatic sentence

Suppose an event-screening system reports that two tracked objects may approach one another at a predicted time. A reader may naturally ask, ‘Will they collide?’ The message may answer a narrower question: given a particular orbit solution, uncertainty model, screening configuration and data vintage, what is the estimated geometry of a close approach? Translation must not smuggle the wider answer into a narrower claim. It must maintain the boundary between a model output, an engineer’s interpretation, an operational policy and an approved action.

This is why an ordinary dictionary is insufficient. The terms conjunction, screening, miss distance, collision probability, covariance, ephemeris, propagation, object catalogue, primary object and secondary object sit within a chain of systems. The chain begins with observations and orbit determination, continues through future-state propagation and conjunction screening, and reaches operational assessment only when the responsible mission team evaluates available evidence. The translation writer should keep that progression visible, because confusing two stages can create false reassurance or false alarm.

A second source of error is documentation that combines prose and structured data. A CDM can have machine-readable fields that should not be translated, human-readable comments that require careful translation, values whose units and frame must be retained, and assumptions whose absence should trigger a query. A language model or translation memory may make the human wording smoother while inadvertently changing a sign, threshold, identifier or technical relationship. Good review therefore compares the source and target at the level of data and implied action rather than solely at sentence level.

The 2026 environment adds a distinct format hazard. CelesTrak’s documentation says the five-digit catalogue range was exhausted on 11 July 2026 and newly catalogued six-digit objects are not available through its traditional TLE output; it publishes OMM-compatible alternatives. Treat this as a provider-specific delivery constraint, not evidence that a space object vanished or a universal claim that all TLE products ceased to exist. Also note CelesTrak’s documented default GP-query output change to CSV on 9 May 2026. A multilingual technical guide that repeats a pre-2026 format assumption can therefore misdirect both readers and software operators.

What the source describes: observations, orbit states, encounters and decisions

An observation records information obtained by a sensor or analysis system at an identified time. An orbit solution estimates state parameters from observations using a model. Propagation predicts where that state may evolve. Screening searches for potentially close pairings. A conjunction event describes a predicted encounter between two objects, sometimes called the primary protected object and the secondary object. A risk assessment interprets the geometry, uncertainty and possible consequences. Mitigation is a separately authorized activity that may follow assessment. These are different semantic verbs and should remain different in translation.

A message does not become more certain because an editor writes it in plain language. ‘Predicted’ must stay predicted, ‘estimated’ must stay estimated, and ‘reported’ must retain its reporter. Even the subject of a sentence matters: ‘the screening identified a close approach’ is not the same as ‘the operators confirmed a dangerous encounter’. The first describes a system-generated finding; the second makes a human decision and risk judgment claim. Technical translation must also preserve intentional distinctions between detected, tracked, catalogued, correlated and identified objects.

For each number, ask five questions: what quantity is it, what unit represents it, what coordinate or reference frame gives it meaning, what epoch or interval applies, and what uncertainty accompanies it? For each decision statement, ask what authority made it, under which procedure, with which revision, and whether it remains active. The most informative translation may need a small parenthetical explanation or a correctly labeled glossary entry, but should not manufacture missing metadata. When the source lacks a crucial detail, carry the absence forward and raise a question rather than inventing a convenient default.

This guide is written to support trained translators, technical writers, multilingual educators, documentation designers and reviewers. It is deliberately non-operational: fictional values and examples cannot be used as spacecraft collision risk thresholds, maneuver timings, flight rules or protected-asset decisions. All case studies are invented so the underlying distinction is teachable without implying access to real mission data. The authority for any operational decision remains the spacecraft’s designated operations and safety process.

Sixty-two specialist mechanisms for preserving conjunction meaning

Each mechanism below treats one recurrent translation problem as a small evidence contract. The aim is not to learn a glossary by heart; it is to understand which distinctions a reader would lose if a value, modifier or role moved incorrectly. The same tests apply when the target is another natural language, a plain-language summary, an accessibility layer or a localized software interface.

1. Primary object

What it means. The primary is the protected asset selected for a screening or collision-risk analysis. A translated report should establish the object type and source authority before introducing a friendly synonym. The operational question is not ‘Does this wording sound technical?’ but ‘Does this wording still describe the same kind of evidence under the same conditions?’ One useful diagnostic is to reconstruct the sentence as a data relationship: which entity, at which time, using which model, is said to have which property? This forces the translator to handle roles and qualifications before polishing the prose.

How meaning fails. Calling the primary the heavier, first discovered or physically leading object changes the role. Consider what a reader could reasonably infer from that mistake: the actor, the point in the lifecycle, or the numerical interpretation may shift without any obvious grammatical error. The remedy is to preserve the evidence-bearing distinction before editing for style. Retain its catalog identifier and explicitly describe the role assigned by the data producer. A translator who cannot establish the source meaning should label the decision pending rather than select a plausible expression from general vocabulary.

Practical check. Verify OBJECT1/OBJECT2 mapping and the issuer’s definition of which object is protected. Write a tiny audit receipt with the source field or phrase, its approved target expression, the source reference, the reviewer and the release status. Read the target alone and ask whether it implies a new action, a new certainty, a new unit or an unsupported comparison. Do not approve this field on vocabulary similarity alone. Mark it verified only when the source-to-target evidence trail supports the same assertion and the same limits.

2. Secondary object

What it means. The secondary is the other tracked object involved in the screened close approach. A translated report should establish the object type and source authority before introducing a friendly synonym. The operational question is not ‘Does this wording sound technical?’ but ‘Does this wording still describe the same kind of evidence under the same conditions?’ Another useful diagnostic is a source-to-target trace: highlight every value and qualifier in the source, then annotate the exact passage that carries it in the target. An unpaired item is a question, not a cosmetic omission.

How meaning fails. Calling it necessarily debris is unsupported; a functioning spacecraft may also be secondary. Consider what a reader could reasonably infer from that mistake: the actor, the point in the lifecycle, or the numerical interpretation may shift without any obvious grammatical error. The remedy is to preserve the evidence-bearing distinction before editing for style. Keep its recorded object type and uncertainty qualifiers distinct from its role. A translator who cannot establish the source meaning should label the decision pending rather than select a plausible expression from general vocabulary.

Practical check. Compare the second object identifier across CDM header, object block and human summary. Write a tiny audit receipt with the source field or phrase, its approved target expression, the source reference, the reviewer and the release status. Read the target alone and ask whether it implies a new action, a new certainty, a new unit or an unsupported comparison. The release test is deliberately narrow: can another competent reviewer recover the same field, relationship, time and state without guessing? If not, return the item for clarification.

3. Object identity

What it means. Catalog IDs, international designators and operator labels answer related but distinct identity questions. A translated report should establish the object type and source authority before introducing a friendly synonym. The operational question is not ‘Does this wording sound technical?’ but ‘Does this wording still describe the same kind of evidence under the same conditions?’ The risk is especially high in mixed documents that contain an immutable machine layer and a human commentary layer. The correct treatment is not to translate everything identically: preserve parser-level tokens while explaining human meaning.

How meaning fails. A familiar mission name may refer to several objects, payloads or orbital bodies. Consider what a reader could reasonably infer from that mistake: the actor, the point in the lifecycle, or the numerical interpretation may shift without any obvious grammatical error. The remedy is to preserve the evidence-bearing distinction before editing for style. Retain stable machine identifiers and treat display names as separately sourced aliases. A translator who cannot establish the source meaning should label the decision pending rather than select a plausible expression from general vocabulary.

Practical check. Perform a two-way identifier lookup; never key a safety record on a loose translated name. Write a tiny audit receipt with the source field or phrase, its approved target expression, the source reference, the reviewer and the release status. Read the target alone and ask whether it implies a new action, a new certainty, a new unit or an unsupported comparison. If the source record is incomplete, preserve that incompleteness. A translation must not become a covert data-enrichment pipeline that fills blanks with plausible assumptions.

4. Six-digit catalog numbers

What it means. Newly catalogued objects may carry six-digit numeric IDs outside conventional five-digit TLE representation. A translated report should establish the object type and source authority before introducing a friendly synonym. The operational question is not ‘Does this wording sound technical?’ but ‘Does this wording still describe the same kind of evidence under the same conditions?’ A strong explanation for a general reader should move from observation to prediction to possible decision, explicitly naming the step being discussed. That sequence helps readers understand why a forecast is not a command.

How meaning fails. A missing legacy TLE row may be mistaken for a missing object or fresh orbital safety evidence. Consider what a reader could reasonably infer from that mistake: the actor, the point in the lifecycle, or the numerical interpretation may shift without any obvious grammatical error. The remedy is to preserve the evidence-bearing distinction before editing for style. State the coverage limit and use supported OMM-compatible delivery for these identifiers. A translator who cannot establish the source meaning should label the decision pending rather than select a plausible expression from general vocabulary.

Practical check. Test 99999 and 100000 as separate schema cases without truncation or alphabetic guesswork. Write a tiny audit receipt with the source field or phrase, its approved target expression, the source reference, the reviewer and the release status. Read the target alone and ask whether it implies a new action, a new certainty, a new unit or an unsupported comparison. This is the distinction between fidelity and fluency. Both are desirable, but fluent wording is not a substitute for a traceable scientific claim.

5. Message creation time

What it means. CREATION_DATE identifies when a message version was assembled, not when closest approach occurs. A translated report should establish the object type and source authority before introducing a friendly synonym. The operational question is not ‘Does this wording sound technical?’ but ‘Does this wording still describe the same kind of evidence under the same conditions?’ When two values are numerically plausible, a copied typo may escape ordinary proofreading. A reviewer should therefore compare identifiers, signs, time references and units as independent evidence, not as part of flowing prose.

How meaning fails. Readers may treat the issuance timestamp as the time at which vehicles meet. Consider what a reader could reasonably infer from that mistake: the actor, the point in the lifecycle, or the numerical interpretation may shift without any obvious grammatical error. The remedy is to preserve the evidence-bearing distinction before editing for style. Give message creation and TCA their own labels, preferably preserving the raw ISO-style date. A translator who cannot establish the source meaning should label the decision pending rather than select a plausible expression from general vocabulary.

Practical check. Reconcile header timestamp, event timestamp and any publication time in a three-column check. Write a tiny audit receipt with the source field or phrase, its approved target expression, the source reference, the reviewer and the release status. Read the target alone and ask whether it implies a new action, a new certainty, a new unit or an unsupported comparison. The intended audience may use different terms, but it must not receive a different event. Explain the concept at the surface while holding its data definition steady.

6. Time of closest approach

What it means. TCA is the modelled time of closest approach for the stated pair and prediction solution. A translated report should establish the object type and source authority before introducing a friendly synonym. The operational question is not ‘Does this wording sound technical?’ but ‘Does this wording still describe the same kind of evidence under the same conditions?’ Plain language can carry exact science if it does not hide the conditional structure. Shorten long clauses, but keep the phrases that specify model, event, likelihood, revision and authority in the final sentence.

How meaning fails. Calling TCA a collision time turns a forecasted geometric minimum into an asserted impact. Consider what a reader could reasonably infer from that mistake: the actor, the point in the lifecycle, or the numerical interpretation may shift without any obvious grammatical error. The remedy is to preserve the evidence-bearing distinction before editing for style. Use ‘predicted time of closest approach’ and preserve time scale and precision. A translator who cannot establish the source meaning should label the decision pending rather than select a plausible expression from general vocabulary.

Practical check. Trace TCA changes between alerts before inferring any improvement or deterioration. Write a tiny audit receipt with the source field or phrase, its approved target expression, the source reference, the reviewer and the release status. Read the target alone and ask whether it implies a new action, a new certainty, a new unit or an unsupported comparison. A real operational desk needs a controlled source, not a memory of an earlier alert. The audit should preserve enough provenance to recover what was known at each decision point.

7. UTC and local time

What it means. UTC may be used as the explicit time reference while a translated explanation adds a local presentation. A translated report should establish the object type and source authority before introducing a friendly synonym. The operational question is not ‘Does this wording sound technical?’ but ‘Does this wording still describe the same kind of evidence under the same conditions?’ A search-result snippet is sometimes the only text a reader sees. The title, summary and excerpt must not make a stronger operational claim than the underlying report, even when a more dramatic phrasing attracts clicks.

How meaning fails. A converted clock with no timezone can appear to denote the original machine timestamp. Consider what a reader could reasonably infer from that mistake: the actor, the point in the lifecycle, or the numerical interpretation may shift without any obvious grammatical error. The remedy is to preserve the evidence-bearing distinction before editing for style. Preserve original UTC; mark any human-facing conversion as an added display aid. A translator who cannot establish the source meaning should label the decision pending rather than select a plausible expression from general vocabulary.

Practical check. Check date rollover, daylight-saving behavior, and the suffix of machine fields. Write a tiny audit receipt with the source field or phrase, its approved target expression, the source reference, the reviewer and the release status. Read the target alone and ask whether it implies a new action, a new certainty, a new unit or an unsupported comparison. Do not approve this field on vocabulary similarity alone. Mark it verified only when the source-to-target evidence trail supports the same assertion and the same limits.

8. Time precision

What it means. A timestamp containing fractional seconds implies a specific representation and should not be rounded casually. A translated report should establish the object type and source authority before introducing a friendly synonym. The operational question is not ‘Does this wording sound technical?’ but ‘Does this wording still describe the same kind of evidence under the same conditions?’ An auditor should ask what action a reader could take after seeing the translated statement. If the target copy invites a different decision from the source, the translation has changed meaning even when the nouns look accurate.

How meaning fails. Truncation can conceal ordering or make two distinct revisions appear simultaneous. Consider what a reader could reasonably infer from that mistake: the actor, the point in the lifecycle, or the numerical interpretation may shift without any obvious grammatical error. The remedy is to preserve the evidence-bearing distinction before editing for style. Retain machine precision and document when presentation rounds for readability. A translator who cannot establish the source meaning should label the decision pending rather than select a plausible expression from general vocabulary.

Practical check. Compare numeric date strings character by character for critical events. Write a tiny audit receipt with the source field or phrase, its approved target expression, the source reference, the reviewer and the release status. Read the target alone and ask whether it implies a new action, a new certainty, a new unit or an unsupported comparison. The release test is deliberately narrow: can another competent reviewer recover the same field, relationship, time and state without guessing? If not, return the item for clarification.

9. Screening start and stop

What it means. Screening intervals define the time window over which the system searched for encounters. A translated report should establish the object type and source authority before introducing a friendly synonym. The operational question is not ‘Does this wording sound technical?’ but ‘Does this wording still describe the same kind of evidence under the same conditions?’ One useful diagnostic is to reconstruct the sentence as a data relationship: which entity, at which time, using which model, is said to have which property? This forces the translator to handle roles and qualifications before polishing the prose.

How meaning fails. A translation may present the screening window as an availability or maneuver window. Consider what a reader could reasonably infer from that mistake: the actor, the point in the lifecycle, or the numerical interpretation may shift without any obvious grammatical error. The remedy is to preserve the evidence-bearing distinction before editing for style. Label it as an analysis window, not a recommended activity period. A translator who cannot establish the source meaning should label the decision pending rather than select a plausible expression from general vocabulary.

Practical check. Check whether start, stop and TCA satisfy the intended relationship in the source record. Write a tiny audit receipt with the source field or phrase, its approved target expression, the source reference, the reviewer and the release status. Read the target alone and ask whether it implies a new action, a new certainty, a new unit or an unsupported comparison. If the source record is incomplete, preserve that incompleteness. A translation must not become a covert data-enrichment pipeline that fills blanks with plausible assumptions.

10. Prediction horizon

What it means. A model forecast has a finite horizon governed by available inputs and assumptions. A translated report should establish the object type and source authority before introducing a friendly synonym. The operational question is not ‘Does this wording sound technical?’ but ‘Does this wording still describe the same kind of evidence under the same conditions?’ Another useful diagnostic is a source-to-target trace: highlight every value and qualifier in the source, then annotate the exact passage that carries it in the target. An unpaired item is a question, not a cosmetic omission.

How meaning fails. Readers may imagine a prediction remains valid indefinitely once it has been published. Consider what a reader could reasonably infer from that mistake: the actor, the point in the lifecycle, or the numerical interpretation may shift without any obvious grammatical error. The remedy is to preserve the evidence-bearing distinction before editing for style. Expose the time-bounded nature of the assessment and its data timestamp. A translator who cannot establish the source meaning should label the decision pending rather than select a plausible expression from general vocabulary.

Practical check. Ask whether a subsequent orbit update, maneuver or new observation superseded the prediction. Write a tiny audit receipt with the source field or phrase, its approved target expression, the source reference, the reviewer and the release status. Read the target alone and ask whether it implies a new action, a new certainty, a new unit or an unsupported comparison. This is the distinction between fidelity and fluency. Both are desirable, but fluent wording is not a substitute for a traceable scientific claim.

11. Miss distance

What it means. Miss distance is a predicted separation metric for a modeled encounter under stated assumptions. A translated report should establish the object type and source authority before introducing a friendly synonym. The operational question is not ‘Does this wording sound technical?’ but ‘Does this wording still describe the same kind of evidence under the same conditions?’ The risk is especially high in mixed documents that contain an immutable machine layer and a human commentary layer. The correct treatment is not to translate everything identically: preserve parser-level tokens while explaining human meaning.

How meaning fails. The popular phrase ‘distance between satellites now’ misplaces the measurement in time. Consider what a reader could reasonably infer from that mistake: the actor, the point in the lifecycle, or the numerical interpretation may shift without any obvious grammatical error. The remedy is to preserve the evidence-bearing distinction before editing for style. Translate as predicted minimum separation or miss distance, not present separation. A translator who cannot establish the source meaning should label the decision pending rather than select a plausible expression from general vocabulary.

Practical check. Verify the units and whether the value is scalar or tied to component geometry. Write a tiny audit receipt with the source field or phrase, its approved target expression, the source reference, the reviewer and the release status. Read the target alone and ask whether it implies a new action, a new certainty, a new unit or an unsupported comparison. The intended audience may use different terms, but it must not receive a different event. Explain the concept at the surface while holding its data definition steady.

12. Relative position

What it means. Relative position components are defined within a specific reference frame. A translated report should establish the object type and source authority before introducing a friendly synonym. The operational question is not ‘Does this wording sound technical?’ but ‘Does this wording still describe the same kind of evidence under the same conditions?’ A strong explanation for a general reader should move from observation to prediction to possible decision, explicitly naming the step being discussed. That sequence helps readers understand why a forecast is not a command.

How meaning fails. The sign of a component can be misread as compass direction or geographic movement. Consider what a reader could reasonably infer from that mistake: the actor, the point in the lifecycle, or the numerical interpretation may shift without any obvious grammatical error. The remedy is to preserve the evidence-bearing distinction before editing for style. Retain signed components and frame labels; explain their axes separately. A translator who cannot establish the source meaning should label the decision pending rather than select a plausible expression from general vocabulary.

Practical check. Check the order of axes and the definition of radial, along-track and cross-track. Write a tiny audit receipt with the source field or phrase, its approved target expression, the source reference, the reviewer and the release status. Read the target alone and ask whether it implies a new action, a new certainty, a new unit or an unsupported comparison. A real operational desk needs a controlled source, not a memory of an earlier alert. The audit should preserve enough provenance to recover what was known at each decision point.

13. Relative velocity

What it means. Relative velocity describes velocity of one object relative to the other at the encounter geometry. A translated report should establish the object type and source authority before introducing a friendly synonym. The operational question is not ‘Does this wording sound technical?’ but ‘Does this wording still describe the same kind of evidence under the same conditions?’ When two values are numerically plausible, a copied typo may escape ordinary proofreading. A reviewer should therefore compare identifiers, signs, time references and units as independent evidence, not as part of flowing prose.

How meaning fails. A nontechnical rendering may confuse it with each object’s absolute orbital speed. Consider what a reader could reasonably infer from that mistake: the actor, the point in the lifecycle, or the numerical interpretation may shift without any obvious grammatical error. The remedy is to preserve the evidence-bearing distinction before editing for style. State the relational nature and the stated unit, without inventing a speed of impact. A translator who cannot establish the source meaning should label the decision pending rather than select a plausible expression from general vocabulary.

Practical check. Compare units and vector/scalar status with the original message. Write a tiny audit receipt with the source field or phrase, its approved target expression, the source reference, the reviewer and the release status. Read the target alone and ask whether it implies a new action, a new certainty, a new unit or an unsupported comparison. Do not approve this field on vocabulary similarity alone. Mark it verified only when the source-to-target evidence trail supports the same assertion and the same limits.

14. Coordinate reference frames

What it means. Frame metadata determine how state vectors and components are interpreted. A translated report should establish the object type and source authority before introducing a friendly synonym. The operational question is not ‘Does this wording sound technical?’ but ‘Does this wording still describe the same kind of evidence under the same conditions?’ Plain language can carry exact science if it does not hide the conditional structure. Shorten long clauses, but keep the phrases that specify model, event, likelihood, revision and authority in the final sentence.

How meaning fails. A numerical triple can be copied faithfully yet become wrong if a translated caption names another frame. Consider what a reader could reasonably infer from that mistake: the actor, the point in the lifecycle, or the numerical interpretation may shift without any obvious grammatical error. The remedy is to preserve the evidence-bearing distinction before editing for style. Treat frame tokens as immutable keys and explain them with a cited frame definition. A translator who cannot establish the source meaning should label the decision pending rather than select a plausible expression from general vocabulary.

Practical check. Check that rotation, origin, epoch and reference-frame assumptions are not mixed. Write a tiny audit receipt with the source field or phrase, its approved target expression, the source reference, the reviewer and the release status. Read the target alone and ask whether it implies a new action, a new certainty, a new unit or an unsupported comparison. The release test is deliberately narrow: can another competent reviewer recover the same field, relationship, time and state without guessing? If not, return the item for clarification.

15. RTN coordinates

What it means. Radial, transverse and normal are geometrically defined local directions in orbital dynamics. A translated report should establish the object type and source authority before introducing a friendly synonym. The operational question is not ‘Does this wording sound technical?’ but ‘Does this wording still describe the same kind of evidence under the same conditions?’ A search-result snippet is sometimes the only text a reader sees. The title, summary and excerpt must not make a stronger operational claim than the underlying report, even when a more dramatic phrasing attracts clicks.

How meaning fails. Transverse may be translated as generic ‘horizontal’ and normal as ordinary ‘usual’. Consider what a reader could reasonably infer from that mistake: the actor, the point in the lifecycle, or the numerical interpretation may shift without any obvious grammatical error. The remedy is to preserve the evidence-bearing distinction before editing for style. Keep RTN names or define precise equivalents, with axes tied to orbit geometry. A translator who cannot establish the source meaning should label the decision pending rather than select a plausible expression from general vocabulary.

Practical check. Inspect the source’s frame convention before comparing values. Write a tiny audit receipt with the source field or phrase, its approved target expression, the source reference, the reviewer and the release status. Read the target alone and ask whether it implies a new action, a new certainty, a new unit or an unsupported comparison. If the source record is incomplete, preserve that incompleteness. A translation must not become a covert data-enrichment pipeline that fills blanks with plausible assumptions.

16. State vector

What it means. Position and velocity components describe a state at a particular epoch. A translated report should establish the object type and source authority before introducing a friendly synonym. The operational question is not ‘Does this wording sound technical?’ but ‘Does this wording still describe the same kind of evidence under the same conditions?’ An auditor should ask what action a reader could take after seeing the translated statement. If the target copy invites a different decision from the source, the translation has changed meaning even when the nouns look accurate.

How meaning fails. A summary that lists the numbers without their epoch implies timeless physical coordinates. Consider what a reader could reasonably infer from that mistake: the actor, the point in the lifecycle, or the numerical interpretation may shift without any obvious grammatical error. The remedy is to preserve the evidence-bearing distinction before editing for style. Bind the six components to their epoch, reference frame and unit system. A translator who cannot establish the source meaning should label the decision pending rather than select a plausible expression from general vocabulary.

Practical check. Validate number of elements, ordering, signedness and time tag. Write a tiny audit receipt with the source field or phrase, its approved target expression, the source reference, the reviewer and the release status. Read the target alone and ask whether it implies a new action, a new certainty, a new unit or an unsupported comparison. This is the distinction between fidelity and fluency. Both are desirable, but fluent wording is not a substitute for a traceable scientific claim.

17. Orbit determination

What it means. Orbit determination uses data and models to estimate an orbit solution. A translated report should establish the object type and source authority before introducing a friendly synonym. The operational question is not ‘Does this wording sound technical?’ but ‘Does this wording still describe the same kind of evidence under the same conditions?’ One useful diagnostic is to reconstruct the sentence as a data relationship: which entity, at which time, using which model, is said to have which property? This forces the translator to handle roles and qualifications before polishing the prose.

How meaning fails. Reporting an estimate as direct observation hides assumptions and residual error. Consider what a reader could reasonably infer from that mistake: the actor, the point in the lifecycle, or the numerical interpretation may shift without any obvious grammatical error. The remedy is to preserve the evidence-bearing distinction before editing for style. Use ‘estimated from observations’ where that is the demonstrated status. A translator who cannot establish the source meaning should label the decision pending rather than select a plausible expression from general vocabulary.

Practical check. Review observation span, data fit indicators and analysis provenance. Write a tiny audit receipt with the source field or phrase, its approved target expression, the source reference, the reviewer and the release status. Read the target alone and ask whether it implies a new action, a new certainty, a new unit or an unsupported comparison. The intended audience may use different terms, but it must not receive a different event. Explain the concept at the surface while holding its data definition steady.

18. Observation span

What it means. The observation interval indicates the history of data contributing to an orbit solution. A translated report should establish the object type and source authority before introducing a friendly synonym. The operational question is not ‘Does this wording sound technical?’ but ‘Does this wording still describe the same kind of evidence under the same conditions?’ Another useful diagnostic is a source-to-target trace: highlight every value and qualifier in the source, then annotate the exact passage that carries it in the target. An unpaired item is a question, not a cosmetic omission.

How meaning fails. A stale orbit may look fresh because a CDM was generated recently. Consider what a reader could reasonably infer from that mistake: the actor, the point in the lifecycle, or the numerical interpretation may shift without any obvious grammatical error. The remedy is to preserve the evidence-bearing distinction before editing for style. Keep data-span endpoints distinct from the timestamp of report generation. A translator who cannot establish the source meaning should label the decision pending rather than select a plausible expression from general vocabulary.

Practical check. Check whether report time is later than the last contributing observation. Write a tiny audit receipt with the source field or phrase, its approved target expression, the source reference, the reviewer and the release status. Read the target alone and ask whether it implies a new action, a new certainty, a new unit or an unsupported comparison. A real operational desk needs a controlled source, not a memory of an earlier alert. The audit should preserve enough provenance to recover what was known at each decision point.

19. Covariance matrix

What it means. Covariance expresses uncertainty and correlations in estimated state components. A translated report should establish the object type and source authority before introducing a friendly synonym. The operational question is not ‘Does this wording sound technical?’ but ‘Does this wording still describe the same kind of evidence under the same conditions?’ The risk is especially high in mixed documents that contain an immutable machine layer and a human commentary layer. The correct treatment is not to translate everything identically: preserve parser-level tokens while explaining human meaning.

How meaning fails. Replacing it with ‘error margin of X metres’ discards orientation, correlation and parameter relationships. Consider what a reader could reasonably infer from that mistake: the actor, the point in the lifecycle, or the numerical interpretation may shift without any obvious grammatical error. The remedy is to preserve the evidence-bearing distinction before editing for style. Describe it as an uncertainty representation; retain numerical matrix and frame. A translator who cannot establish the source meaning should label the decision pending rather than select a plausible expression from general vocabulary.

Practical check. Check symmetry, units, index order and whether reported elements match expected fields. Write a tiny audit receipt with the source field or phrase, its approved target expression, the source reference, the reviewer and the release status. Read the target alone and ask whether it implies a new action, a new certainty, a new unit or an unsupported comparison. Do not approve this field on vocabulary similarity alone. Mark it verified only when the source-to-target evidence trail supports the same assertion and the same limits.

20. Covariance diagonals

What it means. Diagonal covariance entries are variances, not the corresponding one-sigma standard deviations. A translated report should establish the object type and source authority before introducing a friendly synonym. The operational question is not ‘Does this wording sound technical?’ but ‘Does this wording still describe the same kind of evidence under the same conditions?’ A strong explanation for a general reader should move from observation to prediction to possible decision, explicitly naming the step being discussed. That sequence helps readers understand why a forecast is not a command.

How meaning fails. A translator may read a squared unit value as a plain linear uncertainty. Consider what a reader could reasonably infer from that mistake: the actor, the point in the lifecycle, or the numerical interpretation may shift without any obvious grammatical error. The remedy is to preserve the evidence-bearing distinction before editing for style. Avoid unsanctioned square-root conversions; state exactly which quantity is displayed. A translator who cannot establish the source meaning should label the decision pending rather than select a plausible expression from general vocabulary.

Practical check. Inspect the dimensional unit and compare conversion with a qualified analyst if needed. Write a tiny audit receipt with the source field or phrase, its approved target expression, the source reference, the reviewer and the release status. Read the target alone and ask whether it implies a new action, a new certainty, a new unit or an unsupported comparison. The release test is deliberately narrow: can another competent reviewer recover the same field, relationship, time and state without guessing? If not, return the item for clarification.

21. Covariance cross terms

What it means. Off-diagonal covariance terms encode relationships among variables. A translated report should establish the object type and source authority before introducing a friendly synonym. The operational question is not ‘Does this wording sound technical?’ but ‘Does this wording still describe the same kind of evidence under the same conditions?’ When two values are numerically plausible, a copied typo may escape ordinary proofreading. A reviewer should therefore compare identifiers, signs, time references and units as independent evidence, not as part of flowing prose.

How meaning fails. They may be dropped as unimportant extras, changing the uncertainty model. Consider what a reader could reasonably infer from that mistake: the actor, the point in the lifecycle, or the numerical interpretation may shift without any obvious grammatical error. The remedy is to preserve the evidence-bearing distinction before editing for style. Preserve complete field sets and treat signs as data-bearing. A translator who cannot establish the source meaning should label the decision pending rather than select a plausible expression from general vocabulary.

Practical check. Use a matrix checklist rather than checking only the diagonal values. Write a tiny audit receipt with the source field or phrase, its approved target expression, the source reference, the reviewer and the release status. Read the target alone and ask whether it implies a new action, a new certainty, a new unit or an unsupported comparison. If the source record is incomplete, preserve that incompleteness. A translation must not become a covert data-enrichment pipeline that fills blanks with plausible assumptions.

22. Covariance epoch

What it means. An uncertainty model depends on the epoch to which it was estimated or propagated. A translated report should establish the object type and source authority before introducing a friendly synonym. The operational question is not ‘Does this wording sound technical?’ but ‘Does this wording still describe the same kind of evidence under the same conditions?’ Plain language can carry exact science if it does not hide the conditional structure. Shorten long clauses, but keep the phrases that specify model, event, likelihood, revision and authority in the final sentence.

How meaning fails. The right matrix attached to the wrong time is not the same risk evidence. Consider what a reader could reasonably infer from that mistake: the actor, the point in the lifecycle, or the numerical interpretation may shift without any obvious grammatical error. The remedy is to preserve the evidence-bearing distinction before editing for style. Bind uncertainty values to their valid epoch and propagation treatment. A translator who cannot establish the source meaning should label the decision pending rather than select a plausible expression from general vocabulary.

Practical check. Compare the matrix’s time reference with TCA and state-vector metadata. Write a tiny audit receipt with the source field or phrase, its approved target expression, the source reference, the reviewer and the release status. Read the target alone and ask whether it implies a new action, a new certainty, a new unit or an unsupported comparison. This is the distinction between fidelity and fluency. Both are desirable, but fluent wording is not a substitute for a traceable scientific claim.

23. Confidence and distribution

What it means. Probability interpretations depend on statistical and modeling assumptions. A translated report should establish the object type and source authority before introducing a friendly synonym. The operational question is not ‘Does this wording sound technical?’ but ‘Does this wording still describe the same kind of evidence under the same conditions?’ A search-result snippet is sometimes the only text a reader sees. The title, summary and excerpt must not make a stronger operational claim than the underlying report, even when a more dramatic phrasing attracts clicks.

How meaning fails. A percentage can be translated without noting it is conditional on the stated model. Consider what a reader could reasonably infer from that mistake: the actor, the point in the lifecycle, or the numerical interpretation may shift without any obvious grammatical error. The remedy is to preserve the evidence-bearing distinction before editing for style. Describe likelihood estimates with their assumptions and provenance. A translator who cannot establish the source meaning should label the decision pending rather than select a plausible expression from general vocabulary.

Practical check. Reject language implying a physical guarantee from a modeled probabilistic result. Write a tiny audit receipt with the source field or phrase, its approved target expression, the source reference, the reviewer and the release status. Read the target alone and ask whether it implies a new action, a new certainty, a new unit or an unsupported comparison. The intended audience may use different terms, but it must not receive a different event. Explain the concept at the surface while holding its data definition steady.

24. Collision probability

What it means. Probability of collision is an analytical estimate derived from encounter geometry and uncertainty models. A translated report should establish the object type and source authority before introducing a friendly synonym. The operational question is not ‘Does this wording sound technical?’ but ‘Does this wording still describe the same kind of evidence under the same conditions?’ An auditor should ask what action a reader could take after seeing the translated statement. If the target copy invites a different decision from the source, the translation has changed meaning even when the nouns look accurate.

How meaning fails. It is not a statement that an impact will or will not occur, nor necessarily an across-mission rate. Consider what a reader could reasonably infer from that mistake: the actor, the point in the lifecycle, or the numerical interpretation may shift without any obvious grammatical error. The remedy is to preserve the evidence-bearing distinction before editing for style. Use ‘estimated probability of collision for this modeled encounter’ where justified. A translator who cannot establish the source meaning should label the decision pending rather than select a plausible expression from general vocabulary.

Practical check. Compare scientific notation, thresholds and revision trajectory without editorially rounding risk away. Write a tiny audit receipt with the source field or phrase, its approved target expression, the source reference, the reviewer and the release status. Read the target alone and ask whether it implies a new action, a new certainty, a new unit or an unsupported comparison. A real operational desk needs a controlled source, not a memory of an earlier alert. The audit should preserve enough provenance to recover what was known at each decision point.

25. Scientific notation

What it means. Compact numbers such as 2.0E-05 denote precise orders of magnitude. A translated report should establish the object type and source authority before introducing a friendly synonym. The operational question is not ‘Does this wording sound technical?’ but ‘Does this wording still describe the same kind of evidence under the same conditions?’ One useful diagnostic is to reconstruct the sentence as a data relationship: which entity, at which time, using which model, is said to have which property? This forces the translator to handle roles and qualifications before polishing the prose.

How meaning fails. Decimal separators, negative exponents and percent conversion are common loss channels. Consider what a reader could reasonably infer from that mistake: the actor, the point in the lifecycle, or the numerical interpretation may shift without any obvious grammatical error. The remedy is to preserve the evidence-bearing distinction before editing for style. Copy raw numeric representation in technical fields; add clearly marked prose if helpful. A translator who cannot establish the source meaning should label the decision pending rather than select a plausible expression from general vocabulary.

Practical check. Test that 2E-5 remains 0.00002, not 0.002 or 2 percent. Write a tiny audit receipt with the source field or phrase, its approved target expression, the source reference, the reviewer and the release status. Read the target alone and ask whether it implies a new action, a new certainty, a new unit or an unsupported comparison. Do not approve this field on vocabulary similarity alone. Mark it verified only when the source-to-target evidence trail supports the same assertion and the same limits.

26. Hard-body radius

What it means. Hard-body radius is a collision geometry parameter linked to assumed object sizes and modeling. A translated report should establish the object type and source authority before introducing a friendly synonym. The operational question is not ‘Does this wording sound technical?’ but ‘Does this wording still describe the same kind of evidence under the same conditions?’ Another useful diagnostic is a source-to-target trace: highlight every value and qualifier in the source, then annotate the exact passage that carries it in the target. An unpaired item is a question, not a cosmetic omission.

How meaning fails. It must not be treated as a guaranteed physical measurement of either entire vehicle. Consider what a reader could reasonably infer from that mistake: the actor, the point in the lifecycle, or the numerical interpretation may shift without any obvious grammatical error. The remedy is to preserve the evidence-bearing distinction before editing for style. Keep combined-versus-individual radius assumptions explicit when present. A translator who cannot establish the source meaning should label the decision pending rather than select a plausible expression from general vocabulary.

Practical check. Confirm model inputs and refrain from estimating unreported dimensions. Write a tiny audit receipt with the source field or phrase, its approved target expression, the source reference, the reviewer and the release status. Read the target alone and ask whether it implies a new action, a new certainty, a new unit or an unsupported comparison. The release test is deliberately narrow: can another competent reviewer recover the same field, relationship, time and state without guessing? If not, return the item for clarification.

27. Screening volume

What it means. A screening volume is a geometric region used to select events for further evaluation. A translated report should establish the object type and source authority before introducing a friendly synonym. The operational question is not ‘Does this wording sound technical?’ but ‘Does this wording still describe the same kind of evidence under the same conditions?’ The risk is especially high in mixed documents that contain an immutable machine layer and a human commentary layer. The correct treatment is not to translate everything identically: preserve parser-level tokens while explaining human meaning.

How meaning fails. Readers may confuse being within a search volume with exceeding a collision-risk threshold. Consider what a reader could reasonably infer from that mistake: the actor, the point in the lifecycle, or the numerical interpretation may shift without any obvious grammatical error. The remedy is to preserve the evidence-bearing distinction before editing for style. Call it a filter or screening condition and keep subsequent risk analysis separate. A translator who cannot establish the source meaning should label the decision pending rather than select a plausible expression from general vocabulary.

Practical check. Compare screening criteria with risk-assessment and maneuver thresholds. Write a tiny audit receipt with the source field or phrase, its approved target expression, the source reference, the reviewer and the release status. Read the target alone and ask whether it implies a new action, a new certainty, a new unit or an unsupported comparison. If the source record is incomplete, preserve that incompleteness. A translation must not become a covert data-enrichment pipeline that fills blanks with plausible assumptions.

28. Alert severity

What it means. A provider or operator may assign categories using a named local policy. A translated report should establish the object type and source authority before introducing a friendly synonym. The operational question is not ‘Does this wording sound technical?’ but ‘Does this wording still describe the same kind of evidence under the same conditions?’ A strong explanation for a general reader should move from observation to prediction to possible decision, explicitly naming the step being discussed. That sequence helps readers understand why a forecast is not a command.

How meaning fails. Translating severity labels as universal global classes makes a local convention authoritative. Consider what a reader could reasonably infer from that mistake: the actor, the point in the lifecycle, or the numerical interpretation may shift without any obvious grammatical error. The remedy is to preserve the evidence-bearing distinction before editing for style. Keep the category and the issuing scheme together. A translator who cannot establish the source meaning should label the decision pending rather than select a plausible expression from general vocabulary.

Practical check. Check that labels and colors are interpreted only under their stated owner. Write a tiny audit receipt with the source field or phrase, its approved target expression, the source reference, the reviewer and the release status. Read the target alone and ask whether it implies a new action, a new certainty, a new unit or an unsupported comparison. This is the distinction between fidelity and fluency. Both are desirable, but fluent wording is not a substitute for a traceable scientific claim.

29. Event status

What it means. An event can be newly detected, updated, assessed, closed or superseded. A translated report should establish the object type and source authority before introducing a friendly synonym. The operational question is not ‘Does this wording sound technical?’ but ‘Does this wording still describe the same kind of evidence under the same conditions?’ When two values are numerically plausible, a copied typo may escape ordinary proofreading. A reviewer should therefore compare identifiers, signs, time references and units as independent evidence, not as part of flowing prose.

How meaning fails. A summary that only prints a hazard title may imply the original status remains active. Consider what a reader could reasonably infer from that mistake: the actor, the point in the lifecycle, or the numerical interpretation may shift without any obvious grammatical error. The remedy is to preserve the evidence-bearing distinction before editing for style. Publish current status with the relevant record version and date. A translator who cannot establish the source meaning should label the decision pending rather than select a plausible expression from general vocabulary.

Practical check. Look for successor notices and reversal or closure statements. Write a tiny audit receipt with the source field or phrase, its approved target expression, the source reference, the reviewer and the release status. Read the target alone and ask whether it implies a new action, a new certainty, a new unit or an unsupported comparison. The intended audience may use different terms, but it must not receive a different event. Explain the concept at the surface while holding its data definition steady.

30. Message revision

What it means. Successive CDMs can revise positions, covariance, risk and timing for the same encounter. A translated report should establish the object type and source authority before introducing a friendly synonym. The operational question is not ‘Does this wording sound technical?’ but ‘Does this wording still describe the same kind of evidence under the same conditions?’ Plain language can carry exact science if it does not hide the conditional structure. Shorten long clauses, but keep the phrases that specify model, event, likelihood, revision and authority in the final sentence.

How meaning fails. A later CDM may be treated as a duplicate because the object pair did not change. Consider what a reader could reasonably infer from that mistake: the actor, the point in the lifecycle, or the numerical interpretation may shift without any obvious grammatical error. The remedy is to preserve the evidence-bearing distinction before editing for style. Keep message ID, generation time and revision lineage visible. A translator who cannot establish the source meaning should label the decision pending rather than select a plausible expression from general vocabulary.

Practical check. Diff TCA, miss distance, probability and object metadata across message versions. Write a tiny audit receipt with the source field or phrase, its approved target expression, the source reference, the reviewer and the release status. Read the target alone and ask whether it implies a new action, a new certainty, a new unit or an unsupported comparison. A real operational desk needs a controlled source, not a memory of an earlier alert. The audit should preserve enough provenance to recover what was known at each decision point.

31. Issuer and authority

What it means. The originator of a message may be a screening provider rather than the spacecraft’s maneuver authority. A translated report should establish the object type and source authority before introducing a friendly synonym. The operational question is not ‘Does this wording sound technical?’ but ‘Does this wording still describe the same kind of evidence under the same conditions?’ A search-result snippet is sometimes the only text a reader sees. The title, summary and excerpt must not make a stronger operational claim than the underlying report, even when a more dramatic phrasing attracts clicks.

How meaning fails. A translated first-person construction can mistakenly assign the recommendation to mission control. Consider what a reader could reasonably infer from that mistake: the actor, the point in the lifecycle, or the numerical interpretation may shift without any obvious grammatical error. The remedy is to preserve the evidence-bearing distinction before editing for style. Name the actual issuing body and distinguish assessment from authorization. A translator who cannot establish the source meaning should label the decision pending rather than select a plausible expression from general vocabulary.

Practical check. Check header provenance and the procedural role of each named actor. Write a tiny audit receipt with the source field or phrase, its approved target expression, the source reference, the reviewer and the release status. Read the target alone and ask whether it implies a new action, a new certainty, a new unit or an unsupported comparison. Do not approve this field on vocabulary similarity alone. Mark it verified only when the source-to-target evidence trail supports the same assertion and the same limits.

32. Comments and free text

What it means. Human comments are translated prose, unlike most parser field names. A translated report should establish the object type and source authority before introducing a friendly synonym. The operational question is not ‘Does this wording sound technical?’ but ‘Does this wording still describe the same kind of evidence under the same conditions?’ An auditor should ask what action a reader could take after seeing the translated statement. If the target copy invites a different decision from the source, the translation has changed meaning even when the nouns look accurate.

How meaning fails. A generic language engine may sanitize shorthand and unintentionally remove caveats. Consider what a reader could reasonably infer from that mistake: the actor, the point in the lifecycle, or the numerical interpretation may shift without any obvious grammatical error. The remedy is to preserve the evidence-bearing distinction before editing for style. Translate comments in context while preserving quoted keys and units unchanged. A translator who cannot establish the source meaning should label the decision pending rather than select a plausible expression from general vocabulary.

Practical check. Cross-check every qualifier in free text against structured fields. Write a tiny audit receipt with the source field or phrase, its approved target expression, the source reference, the reviewer and the release status. Read the target alone and ask whether it implies a new action, a new certainty, a new unit or an unsupported comparison. The release test is deliberately narrow: can another competent reviewer recover the same field, relationship, time and state without guessing? If not, return the item for clarification.

33. Machine-readable field names

What it means. Keywords in CDM, OMM, XML, JSON or KVN are part of a data schema. A translated report should establish the object type and source authority before introducing a friendly synonym. The operational question is not ‘Does this wording sound technical?’ but ‘Does this wording still describe the same kind of evidence under the same conditions?’ One useful diagnostic is to reconstruct the sentence as a data relationship: which entity, at which time, using which model, is said to have which property? This forces the translator to handle roles and qualifications before polishing the prose.

How meaning fails. Translating a key such as OBJECT_ID or TCA can break a parser or mapping. Consider what a reader could reasonably infer from that mistake: the actor, the point in the lifecycle, or the numerical interpretation may shift without any obvious grammatical error. The remedy is to preserve the evidence-bearing distinction before editing for style. Localise documentation labels, never protocol tokens inside structured payloads. A translator who cannot establish the source meaning should label the decision pending rather than select a plausible expression from general vocabulary.

Practical check. Run schema validation and compare a machine-readable sample before and after. Write a tiny audit receipt with the source field or phrase, its approved target expression, the source reference, the reviewer and the release status. Read the target alone and ask whether it implies a new action, a new certainty, a new unit or an unsupported comparison. If the source record is incomplete, preserve that incompleteness. A translation must not become a covert data-enrichment pipeline that fills blanks with plausible assumptions.

34. Missing data

What it means. An absent probability or missing covariance should not be silently treated as zero. A translated report should establish the object type and source authority before introducing a friendly synonym. The operational question is not ‘Does this wording sound technical?’ but ‘Does this wording still describe the same kind of evidence under the same conditions?’ Another useful diagnostic is a source-to-target trace: highlight every value and qualifier in the source, then annotate the exact passage that carries it in the target. An unpaired item is a question, not a cosmetic omission.

How meaning fails. Readers may read an empty field as ‘no collision risk’ rather than ‘not provided’. Consider what a reader could reasonably infer from that mistake: the actor, the point in the lifecycle, or the numerical interpretation may shift without any obvious grammatical error. The remedy is to preserve the evidence-bearing distinction before editing for style. Represent missing as unavailable or not reported, consistent with the source. A translator who cannot establish the source meaning should label the decision pending rather than select a plausible expression from general vocabulary.

Practical check. Audit null, omitted and zero as distinct data states. Write a tiny audit receipt with the source field or phrase, its approved target expression, the source reference, the reviewer and the release status. Read the target alone and ask whether it implies a new action, a new certainty, a new unit or an unsupported comparison. This is the distinction between fidelity and fluency. Both are desirable, but fluent wording is not a substitute for a traceable scientific claim.

35. Conflicting sources

What it means. Two providers may report distinct event estimates from different data vintages or methods. A translated report should establish the object type and source authority before introducing a friendly synonym. The operational question is not ‘Does this wording sound technical?’ but ‘Does this wording still describe the same kind of evidence under the same conditions?’ The risk is especially high in mixed documents that contain an immutable machine layer and a human commentary layer. The correct treatment is not to translate everything identically: preserve parser-level tokens while explaining human meaning.

How meaning fails. Combining a time from one source and probability from another can create a synthetic record. Consider what a reader could reasonably infer from that mistake: the actor, the point in the lifecycle, or the numerical interpretation may shift without any obvious grammatical error. The remedy is to preserve the evidence-bearing distinction before editing for style. Preserve source ownership and time of every value; compare without splicing. A translator who cannot establish the source meaning should label the decision pending rather than select a plausible expression from general vocabulary.

Practical check. Keep a row-wise provenance ledger across providers and update cycles. Write a tiny audit receipt with the source field or phrase, its approved target expression, the source reference, the reviewer and the release status. Read the target alone and ask whether it implies a new action, a new certainty, a new unit or an unsupported comparison. The intended audience may use different terms, but it must not receive a different event. Explain the concept at the surface while holding its data definition steady.

36. Orbit Mean-Elements Message

What it means. OMM represents orbital element records in a standards-based form useful beyond traditional TLE fields. A translated report should establish the object type and source authority before introducing a friendly synonym. The operational question is not ‘Does this wording sound technical?’ but ‘Does this wording still describe the same kind of evidence under the same conditions?’ A strong explanation for a general reader should move from observation to prediction to possible decision, explicitly naming the step being discussed. That sequence helps readers understand why a forecast is not a command.

How meaning fails. Describing OMM as merely a longer TLE may hide schema, metadata and representation changes. Consider what a reader could reasonably infer from that mistake: the actor, the point in the lifecycle, or the numerical interpretation may shift without any obvious grammatical error. The remedy is to preserve the evidence-bearing distinction before editing for style. Explain that it preserves machine-readable keywords and data-context requirements. A translator who cannot establish the source meaning should label the decision pending rather than select a plausible expression from general vocabulary.

Practical check. Validate object ID width, epoch, element theory and serialization format. Write a tiny audit receipt with the source field or phrase, its approved target expression, the source reference, the reviewer and the release status. Read the target alone and ask whether it implies a new action, a new certainty, a new unit or an unsupported comparison. A real operational desk needs a controlled source, not a memory of an earlier alert. The audit should preserve enough provenance to recover what was known at each decision point.

37. TLE format

What it means. The traditional two-line element format has fixed-width conventions and historic identifier limits. A translated report should establish the object type and source authority before introducing a friendly synonym. The operational question is not ‘Does this wording sound technical?’ but ‘Does this wording still describe the same kind of evidence under the same conditions?’ When two values are numerically plausible, a copied typo may escape ordinary proofreading. A reviewer should therefore compare identifiers, signs, time references and units as independent evidence, not as part of flowing prose.

How meaning fails. An unchanged TLE parser may silently omit modern six-digit objects. Consider what a reader could reasonably infer from that mistake: the actor, the point in the lifecycle, or the numerical interpretation may shift without any obvious grammatical error. The remedy is to preserve the evidence-bearing distinction before editing for style. Label legacy limitations, support alternatives where needed and avoid claims of universal TLE coverage. A translator who cannot establish the source meaning should label the decision pending rather than select a plausible expression from general vocabulary.

Practical check. Run boundary tests for 5-digit and 6-digit catalogue identifiers. Write a tiny audit receipt with the source field or phrase, its approved target expression, the source reference, the reviewer and the release status. Read the target alone and ask whether it implies a new action, a new certainty, a new unit or an unsupported comparison. Do not approve this field on vocabulary similarity alone. Mark it verified only when the source-to-target evidence trail supports the same assertion and the same limits.

38. Default GP output

What it means. A provider’s unspecified query format can change independently of an application’s expectations. A translated report should establish the object type and source authority before introducing a friendly synonym. The operational question is not ‘Does this wording sound technical?’ but ‘Does this wording still describe the same kind of evidence under the same conditions?’ Plain language can carry exact science if it does not hide the conditional structure. Shorten long clauses, but keep the phrases that specify model, event, likelihood, revision and authority in the final sentence.

How meaning fails. A response assumed to be TLE may arrive as CSV, causing a parser or documentation error. Consider what a reader could reasonably infer from that mistake: the actor, the point in the lifecycle, or the numerical interpretation may shift without any obvious grammatical error. The remedy is to preserve the evidence-bearing distinction before editing for style. Always request the desired supported format explicitly and document the query. A translator who cannot establish the source meaning should label the decision pending rather than select a plausible expression from general vocabulary.

Practical check. Test content type and schema rather than infer from URL suffix. Write a tiny audit receipt with the source field or phrase, its approved target expression, the source reference, the reviewer and the release status. Read the target alone and ask whether it implies a new action, a new certainty, a new unit or an unsupported comparison. The release test is deliberately narrow: can another competent reviewer recover the same field, relationship, time and state without guessing? If not, return the item for clarification.

39. SGP4 and mean elements

What it means. General perturbations data are meant for appropriate propagation approaches, not arbitrary classical-element arithmetic. A translated report should establish the object type and source authority before introducing a friendly synonym. The operational question is not ‘Does this wording sound technical?’ but ‘Does this wording still describe the same kind of evidence under the same conditions?’ A search-result snippet is sometimes the only text a reader sees. The title, summary and excerpt must not make a stronger operational claim than the underlying report, even when a more dramatic phrasing attracts clicks.

How meaning fails. A prose guide that invites incompatible calculations can turn terminology advice into wrong dynamics. Consider what a reader could reasonably infer from that mistake: the actor, the point in the lifecycle, or the numerical interpretation may shift without any obvious grammatical error. The remedy is to preserve the evidence-bearing distinction before editing for style. Name the model context without supplying operational prediction recipes. A translator who cannot establish the source meaning should label the decision pending rather than select a plausible expression from general vocabulary.

Practical check. Check field meanings against authoritative format and algorithm documentation. Write a tiny audit receipt with the source field or phrase, its approved target expression, the source reference, the reviewer and the release status. Read the target alone and ask whether it implies a new action, a new certainty, a new unit or an unsupported comparison. If the source record is incomplete, preserve that incompleteness. A translation must not become a covert data-enrichment pipeline that fills blanks with plausible assumptions.

40. Epoch and validity

What it means. An element set’s epoch identifies reference time for the orbit state. A translated report should establish the object type and source authority before introducing a friendly synonym. The operational question is not ‘Does this wording sound technical?’ but ‘Does this wording still describe the same kind of evidence under the same conditions?’ An auditor should ask what action a reader could take after seeing the translated statement. If the target copy invites a different decision from the source, the translation has changed meaning even when the nouns look accurate.

How meaning fails. The time printed nearest an element set may be mistaken for the date the data were downloaded. Consider what a reader could reasonably infer from that mistake: the actor, the point in the lifecycle, or the numerical interpretation may shift without any obvious grammatical error. The remedy is to preserve the evidence-bearing distinction before editing for style. Distinguish epoch from retrieval, production and import timestamps. A translator who cannot establish the source meaning should label the decision pending rather than select a plausible expression from general vocabulary.

Practical check. Confirm time-scale semantics and propagate none of the implied precision into unsupported forecasts. Write a tiny audit receipt with the source field or phrase, its approved target expression, the source reference, the reviewer and the release status. Read the target alone and ask whether it implies a new action, a new certainty, a new unit or an unsupported comparison. This is the distinction between fidelity and fluency. Both are desirable, but fluent wording is not a substitute for a traceable scientific claim.

41. Orbit change and maneuver

What it means. An executed maneuver changes the information required for subsequent orbit predictions. A translated report should establish the object type and source authority before introducing a friendly synonym. The operational question is not ‘Does this wording sound technical?’ but ‘Does this wording still describe the same kind of evidence under the same conditions?’ One useful diagnostic is to reconstruct the sentence as a data relationship: which entity, at which time, using which model, is said to have which property? This forces the translator to handle roles and qualifications before polishing the prose.

How meaning fails. A forecast generated before a burn may be presented as if it incorporates the new trajectory. Consider what a reader could reasonably infer from that mistake: the actor, the point in the lifecycle, or the numerical interpretation may shift without any obvious grammatical error. The remedy is to preserve the evidence-bearing distinction before editing for style. Describe planned, commanded, executed and verified maneuvers as separate states. A translator who cannot establish the source meaning should label the decision pending rather than select a plausible expression from general vocabulary.

Practical check. Require mission-owned post-maneuver evidence before describing the resulting trajectory. Write a tiny audit receipt with the source field or phrase, its approved target expression, the source reference, the reviewer and the release status. Read the target alone and ask whether it implies a new action, a new certainty, a new unit or an unsupported comparison. The intended audience may use different terms, but it must not receive a different event. Explain the concept at the surface while holding its data definition steady.

42. Maneuver recommendation

What it means. A recommendation is a proposed action under a policy, not proof that the spacecraft will move. A translated report should establish the object type and source authority before introducing a friendly synonym. The operational question is not ‘Does this wording sound technical?’ but ‘Does this wording still describe the same kind of evidence under the same conditions?’ Another useful diagnostic is a source-to-target trace: highlight every value and qualifier in the source, then annotate the exact passage that carries it in the target. An unpaired item is a question, not a cosmetic omission.

How meaning fails. Translations using an imperative can convert advisory language into a direct instruction. Consider what a reader could reasonably infer from that mistake: the actor, the point in the lifecycle, or the numerical interpretation may shift without any obvious grammatical error. The remedy is to preserve the evidence-bearing distinction before editing for style. Preserve modal verbs, conditions and approval gates. A translator who cannot establish the source meaning should label the decision pending rather than select a plausible expression from general vocabulary.

Practical check. Check the authority chain and ensure no new command syntax appears in prose. Write a tiny audit receipt with the source field or phrase, its approved target expression, the source reference, the reviewer and the release status. Read the target alone and ask whether it implies a new action, a new certainty, a new unit or an unsupported comparison. A real operational desk needs a controlled source, not a memory of an earlier alert. The audit should preserve enough provenance to recover what was known at each decision point.

43. Maneuver approval

What it means. Approval is an operational status requiring an identifiable decision-maker and scope. A translated report should establish the object type and source authority before introducing a friendly synonym. The operational question is not ‘Does this wording sound technical?’ but ‘Does this wording still describe the same kind of evidence under the same conditions?’ The risk is especially high in mixed documents that contain an immutable machine layer and a human commentary layer. The correct treatment is not to translate everything identically: preserve parser-level tokens while explaining human meaning.

How meaning fails. A comment that a maneuver is ‘under consideration’ may become ‘approved’. Consider what a reader could reasonably infer from that mistake: the actor, the point in the lifecycle, or the numerical interpretation may shift without any obvious grammatical error. The remedy is to preserve the evidence-bearing distinction before editing for style. Retain approval state, sign-off date and constraints if they are actually present. A translator who cannot establish the source meaning should label the decision pending rather than select a plausible expression from general vocabulary.

Practical check. Separate decision records from observations and recommendations. Write a tiny audit receipt with the source field or phrase, its approved target expression, the source reference, the reviewer and the release status. Read the target alone and ask whether it implies a new action, a new certainty, a new unit or an unsupported comparison. Do not approve this field on vocabulary similarity alone. Mark it verified only when the source-to-target evidence trail supports the same assertion and the same limits.

44. Collision versus conjunction

What it means. A conjunction is a modeled close approach, while a collision is actual contact. A translated report should establish the object type and source authority before introducing a friendly synonym. The operational question is not ‘Does this wording sound technical?’ but ‘Does this wording still describe the same kind of evidence under the same conditions?’ A strong explanation for a general reader should move from observation to prediction to possible decision, explicitly naming the step being discussed. That sequence helps readers understand why a forecast is not a command.

How meaning fails. Equating the two inflates the threat or falsely reports an event. Consider what a reader could reasonably infer from that mistake: the actor, the point in the lifecycle, or the numerical interpretation may shift without any obvious grammatical error. The remedy is to preserve the evidence-bearing distinction before editing for style. Use ‘close approach’ for conjunction when writing plain-language introductions. A translator who cannot establish the source meaning should label the decision pending rather than select a plausible expression from general vocabulary.

Practical check. Check every occurrence of collide, impact, near miss and possible encounter. Write a tiny audit receipt with the source field or phrase, its approved target expression, the source reference, the reviewer and the release status. Read the target alone and ask whether it implies a new action, a new certainty, a new unit or an unsupported comparison. The release test is deliberately narrow: can another competent reviewer recover the same field, relationship, time and state without guessing? If not, return the item for clarification.

45. Uncertainty growth

What it means. Prediction uncertainty may change with forecast horizon, observations and modeling choices. A translated report should establish the object type and source authority before introducing a friendly synonym. The operational question is not ‘Does this wording sound technical?’ but ‘Does this wording still describe the same kind of evidence under the same conditions?’ When two values are numerically plausible, a copied typo may escape ordinary proofreading. A reviewer should therefore compare identifiers, signs, time references and units as independent evidence, not as part of flowing prose.

How meaning fails. Readers can mistake a later increase in risk metric for a physical approach getting closer. Consider what a reader could reasonably infer from that mistake: the actor, the point in the lifecycle, or the numerical interpretation may shift without any obvious grammatical error. The remedy is to preserve the evidence-bearing distinction before editing for style. Explain what changed: estimated geometry, uncertainty, input coverage or policy classification. A translator who cannot establish the source meaning should label the decision pending rather than select a plausible expression from general vocabulary.

Practical check. Compare values and assumptions across the same versioned event chain. Write a tiny audit receipt with the source field or phrase, its approved target expression, the source reference, the reviewer and the release status. Read the target alone and ask whether it implies a new action, a new certainty, a new unit or an unsupported comparison. If the source record is incomplete, preserve that incompleteness. A translation must not become a covert data-enrichment pipeline that fills blanks with plausible assumptions.

46. Rounding and precision

What it means. Rounded values are often useful in public copy but unsafe without a visible relationship to raw data. A translated report should establish the object type and source authority before introducing a friendly synonym. The operational question is not ‘Does this wording sound technical?’ but ‘Does this wording still describe the same kind of evidence under the same conditions?’ Plain language can carry exact science if it does not hide the conditional structure. Shorten long clauses, but keep the phrases that specify model, event, likelihood, revision and authority in the final sentence.

How meaning fails. A small probability may round to zero or two distinct times may collapse into one minute. Consider what a reader could reasonably infer from that mistake: the actor, the point in the lifecycle, or the numerical interpretation may shift without any obvious grammatical error. The remedy is to preserve the evidence-bearing distinction before editing for style. Retain raw technical quantities and label display rounding as editorial. A translator who cannot establish the source meaning should label the decision pending rather than select a plausible expression from general vocabulary.

Practical check. Apply minimum precision rules tied to decisions, not arbitrary copy style. Write a tiny audit receipt with the source field or phrase, its approved target expression, the source reference, the reviewer and the release status. Read the target alone and ask whether it implies a new action, a new certainty, a new unit or an unsupported comparison. This is the distinction between fidelity and fluency. Both are desirable, but fluent wording is not a substitute for a traceable scientific claim.

47. Units and dimensions

What it means. Distance, velocity, mass, area and covariance components may employ differing units. A translated report should establish the object type and source authority before introducing a friendly synonym. The operational question is not ‘Does this wording sound technical?’ but ‘Does this wording still describe the same kind of evidence under the same conditions?’ A search-result snippet is sometimes the only text a reader sees. The title, summary and excerpt must not make a stronger operational claim than the underlying report, even when a more dramatic phrasing attracts clicks.

How meaning fails. A translated figure lacking its unit can become a number with no physical meaning. Consider what a reader could reasonably infer from that mistake: the actor, the point in the lifecycle, or the numerical interpretation may shift without any obvious grammatical error. The remedy is to preserve the evidence-bearing distinction before editing for style. Preserve units alongside values; prohibit casual conversion of covariance dimensions. A translator who cannot establish the source meaning should label the decision pending rather than select a plausible expression from general vocabulary.

Practical check. Review both magnitudes and dimensional consistency across the record. Write a tiny audit receipt with the source field or phrase, its approved target expression, the source reference, the reviewer and the release status. Read the target alone and ask whether it implies a new action, a new certainty, a new unit or an unsupported comparison. The intended audience may use different terms, but it must not receive a different event. Explain the concept at the surface while holding its data definition steady.

48. Negative signs

What it means. Signs encode direction in components and exponents, not just typography. A translated report should establish the object type and source authority before introducing a friendly synonym. The operational question is not ‘Does this wording sound technical?’ but ‘Does this wording still describe the same kind of evidence under the same conditions?’ An auditor should ask what action a reader could take after seeing the translated statement. If the target copy invites a different decision from the source, the translation has changed meaning even when the nouns look accurate.

How meaning fails. A dropped minus sign may reverse geometry or expand a tiny probability by orders of magnitude. Consider what a reader could reasonably infer from that mistake: the actor, the point in the lifecycle, or the numerical interpretation may shift without any obvious grammatical error. The remedy is to preserve the evidence-bearing distinction before editing for style. Audit signs as data and use monospaced formatting for raw values. A translator who cannot establish the source meaning should label the decision pending rather than select a plausible expression from general vocabulary.

Practical check. Compare codepoint-level tokens where copying could alter a Unicode minus. Write a tiny audit receipt with the source field or phrase, its approved target expression, the source reference, the reviewer and the release status. Read the target alone and ask whether it implies a new action, a new certainty, a new unit or an unsupported comparison. A real operational desk needs a controlled source, not a memory of an earlier alert. The audit should preserve enough provenance to recover what was known at each decision point.

49. Object classification

What it means. Tracked objects may be payloads, rocket bodies, fragments or uncertain types depending on the catalogue. A translated report should establish the object type and source authority before introducing a friendly synonym. The operational question is not ‘Does this wording sound technical?’ but ‘Does this wording still describe the same kind of evidence under the same conditions?’ One useful diagnostic is to reconstruct the sentence as a data relationship: which entity, at which time, using which model, is said to have which property? This forces the translator to handle roles and qualifications before polishing the prose.

How meaning fails. Calling every secondary object ‘space junk’ can be factually incorrect and stigmatizes legitimate spacecraft. Consider what a reader could reasonably infer from that mistake: the actor, the point in the lifecycle, or the numerical interpretation may shift without any obvious grammatical error. The remedy is to preserve the evidence-bearing distinction before editing for style. Report classification as sourced, and state uncertainty explicitly. A translator who cannot establish the source meaning should label the decision pending rather than select a plausible expression from general vocabulary.

Practical check. Confirm object class, whether active or inactive, and the classification timestamp. Write a tiny audit receipt with the source field or phrase, its approved target expression, the source reference, the reviewer and the release status. Read the target alone and ask whether it implies a new action, a new certainty, a new unit or an unsupported comparison. Do not approve this field on vocabulary similarity alone. Mark it verified only when the source-to-target evidence trail supports the same assertion and the same limits.

50. Public risk statements

What it means. An external audience may need a lay explanation rather than raw CDM fields. A translated report should establish the object type and source authority before introducing a friendly synonym. The operational question is not ‘Does this wording sound technical?’ but ‘Does this wording still describe the same kind of evidence under the same conditions?’ Another useful diagnostic is a source-to-target trace: highlight every value and qualifier in the source, then annotate the exact passage that carries it in the target. An unpaired item is a question, not a cosmetic omission.

How meaning fails. Sensational wording can create panic while jargon can conceal the real conditional risk. Consider what a reader could reasonably infer from that mistake: the actor, the point in the lifecycle, or the numerical interpretation may shift without any obvious grammatical error. The remedy is to preserve the evidence-bearing distinction before editing for style. Explain screening, uncertainty and next review without promising safety or impact. A translator who cannot establish the source meaning should label the decision pending rather than select a plausible expression from general vocabulary.

Practical check. Have both a technical reviewer and representative non-specialist check the same factual claims. Write a tiny audit receipt with the source field or phrase, its approved target expression, the source reference, the reviewer and the release status. Read the target alone and ask whether it implies a new action, a new certainty, a new unit or an unsupported comparison. The release test is deliberately narrow: can another competent reviewer recover the same field, relationship, time and state without guessing? If not, return the item for clarification.

51. Translation memory

What it means. Previously approved sentences can improve consistency when terminology is stable. A translated report should establish the object type and source authority before introducing a friendly synonym. The operational question is not ‘Does this wording sound technical?’ but ‘Does this wording still describe the same kind of evidence under the same conditions?’ The risk is especially high in mixed documents that contain an immutable machine layer and a human commentary layer. The correct treatment is not to translate everything identically: preserve parser-level tokens while explaining human meaning.

How meaning fails. A cached sentence about an earlier event can carry an obsolete time, ID or policy level. Consider what a reader could reasonably infer from that mistake: the actor, the point in the lifecycle, or the numerical interpretation may shift without any obvious grammatical error. The remedy is to preserve the evidence-bearing distinction before editing for style. Parameterize variables, block stale examples and require fresh record comparison. A translator who cannot establish the source meaning should label the decision pending rather than select a plausible expression from general vocabulary.

Practical check. Stress-test reused segments with changed signs, times and status categories. Write a tiny audit receipt with the source field or phrase, its approved target expression, the source reference, the reviewer and the release status. Read the target alone and ask whether it implies a new action, a new certainty, a new unit or an unsupported comparison. If the source record is incomplete, preserve that incompleteness. A translation must not become a covert data-enrichment pipeline that fills blanks with plausible assumptions.

52. Machine translation

What it means. Automated suggestions can help draft prose but may not reliably handle operational distinctions. A translated report should establish the object type and source authority before introducing a friendly synonym. The operational question is not ‘Does this wording sound technical?’ but ‘Does this wording still describe the same kind of evidence under the same conditions?’ A strong explanation for a general reader should move from observation to prediction to possible decision, explicitly naming the step being discussed. That sequence helps readers understand why a forecast is not a command.

How meaning fails. A fluent machine sentence can switch possibility to certainty or drop negation. Consider what a reader could reasonably infer from that mistake: the actor, the point in the lifecycle, or the numerical interpretation may shift without any obvious grammatical error. The remedy is to preserve the evidence-bearing distinction before editing for style. Use bilingual review plus structured invariant tests; never auto-approve risk alerts. A translator who cannot establish the source meaning should label the decision pending rather than select a plausible expression from general vocabulary.

Practical check. Record every critical field’s source-target match and assign an accountable reviewer. Write a tiny audit receipt with the source field or phrase, its approved target expression, the source reference, the reviewer and the release status. Read the target alone and ask whether it implies a new action, a new certainty, a new unit or an unsupported comparison. This is the distinction between fidelity and fluency. Both are desirable, but fluent wording is not a substitute for a traceable scientific claim.

53. Terminology database

What it means. Approved terms should be stored with definitions, context, status and source. A translated report should establish the object type and source authority before introducing a friendly synonym. The operational question is not ‘Does this wording sound technical?’ but ‘Does this wording still describe the same kind of evidence under the same conditions?’ When two values are numerically plausible, a copied typo may escape ordinary proofreading. A reviewer should therefore compare identifiers, signs, time references and units as independent evidence, not as part of flowing prose.

How meaning fails. An isolated glossary translation may choose one target word for several scientific concepts. Consider what a reader could reasonably infer from that mistake: the actor, the point in the lifecycle, or the numerical interpretation may shift without any obvious grammatical error. The remedy is to preserve the evidence-bearing distinction before editing for style. Store concept IDs and forbidden equivalents, not only language pairs. A translator who cannot establish the source meaning should label the decision pending rather than select a plausible expression from general vocabulary.

Practical check. Test examples at sentence level and record disputed terms as unresolved. Write a tiny audit receipt with the source field or phrase, its approved target expression, the source reference, the reviewer and the release status. Read the target alone and ask whether it implies a new action, a new certainty, a new unit or an unsupported comparison. The intended audience may use different terms, but it must not receive a different event. Explain the concept at the surface while holding its data definition steady.

54. Language-specific number formatting

What it means. Different locales vary in decimal and thousands separators. A translated report should establish the object type and source authority before introducing a friendly synonym. The operational question is not ‘Does this wording sound technical?’ but ‘Does this wording still describe the same kind of evidence under the same conditions?’ Plain language can carry exact science if it does not hide the conditional structure. Shorten long clauses, but keep the phrases that specify model, event, likelihood, revision and authority in the final sentence.

How meaning fails. A number like 1,234 or 0,001 can be read as a different quantity. Consider what a reader could reasonably infer from that mistake: the actor, the point in the lifecycle, or the numerical interpretation may shift without any obvious grammatical error. The remedy is to preserve the evidence-bearing distinction before editing for style. Keep machine fields locale-neutral and label target-language numeric presentation. A translator who cannot establish the source meaning should label the decision pending rather than select a plausible expression from general vocabulary.

Practical check. Run roundtrip conversion checks for every safety-relevant numeric field. Write a tiny audit receipt with the source field or phrase, its approved target expression, the source reference, the reviewer and the release status. Read the target alone and ask whether it implies a new action, a new certainty, a new unit or an unsupported comparison. A real operational desk needs a controlled source, not a memory of an earlier alert. The audit should preserve enough provenance to recover what was known at each decision point.

55. Right-to-left display

What it means. Bi-directional text can reorder punctuation around tokens, scientific notation and identifiers. A translated report should establish the object type and source authority before introducing a friendly synonym. The operational question is not ‘Does this wording sound technical?’ but ‘Does this wording still describe the same kind of evidence under the same conditions?’ A search-result snippet is sometimes the only text a reader sees. The title, summary and excerpt must not make a stronger operational claim than the underlying report, even when a more dramatic phrasing attracts clicks.

How meaning fails. The visual reading order may not reflect the underlying string bytes. Consider what a reader could reasonably infer from that mistake: the actor, the point in the lifecycle, or the numerical interpretation may shift without any obvious grammatical error. The remedy is to preserve the evidence-bearing distinction before editing for style. Isolate Latin-script IDs and units using robust content formatting, not manual character reordering. A translator who cannot establish the source meaning should label the decision pending rather than select a plausible expression from general vocabulary.

Practical check. Review rendered and copied versions on representative devices. Write a tiny audit receipt with the source field or phrase, its approved target expression, the source reference, the reviewer and the release status. Read the target alone and ask whether it implies a new action, a new certainty, a new unit or an unsupported comparison. Do not approve this field on vocabulary similarity alone. Mark it verified only when the source-to-target evidence trail supports the same assertion and the same limits.

56. Alt text and accessibility

What it means. Diagrams need accessible explanations of geometry and uncertainty. A translated report should establish the object type and source authority before introducing a friendly synonym. The operational question is not ‘Does this wording sound technical?’ but ‘Does this wording still describe the same kind of evidence under the same conditions?’ An auditor should ask what action a reader could take after seeing the translated statement. If the target copy invites a different decision from the source, the translation has changed meaning even when the nouns look accurate.

How meaning fails. A generic ‘two satellites almost collide’ alt description may assert more than the figure supports. Consider what a reader could reasonably infer from that mistake: the actor, the point in the lifecycle, or the numerical interpretation may shift without any obvious grammatical error. The remedy is to preserve the evidence-bearing distinction before editing for style. Describe axes, time, representation and the fact that positions are predictions. A translator who cannot establish the source meaning should label the decision pending rather than select a plausible expression from general vocabulary.

Practical check. Ask whether a reader relying on text obtains the same caveat as a sighted reader. Write a tiny audit receipt with the source field or phrase, its approved target expression, the source reference, the reviewer and the release status. Read the target alone and ask whether it implies a new action, a new certainty, a new unit or an unsupported comparison. The release test is deliberately narrow: can another competent reviewer recover the same field, relationship, time and state without guessing? If not, return the item for clarification.

57. Screenshots and charts

What it means. Labels and legends can be part of a safety-relevant graphic. A translated report should establish the object type and source authority before introducing a friendly synonym. The operational question is not ‘Does this wording sound technical?’ but ‘Does this wording still describe the same kind of evidence under the same conditions?’ One useful diagnostic is to reconstruct the sentence as a data relationship: which entity, at which time, using which model, is said to have which property? This forces the translator to handle roles and qualifications before polishing the prose.

How meaning fails. Translation overprinted on a diagram can obscure a threshold line or flip a legend. Consider what a reader could reasonably infer from that mistake: the actor, the point in the lifecycle, or the numerical interpretation may shift without any obvious grammatical error. The remedy is to preserve the evidence-bearing distinction before editing for style. Version the visual and independently verify tick marks, units and object mapping. A translator who cannot establish the source meaning should label the decision pending rather than select a plausible expression from general vocabulary.

Practical check. Compare source and target image against the underlying table. Write a tiny audit receipt with the source field or phrase, its approved target expression, the source reference, the reviewer and the release status. Read the target alone and ask whether it implies a new action, a new certainty, a new unit or an unsupported comparison. If the source record is incomplete, preserve that incompleteness. A translation must not become a covert data-enrichment pipeline that fills blanks with plausible assumptions.

58. Incident chronology

What it means. A retrospective record may contain initial estimates, revisions, decisions and outcome. A translated report should establish the object type and source authority before introducing a friendly synonym. The operational question is not ‘Does this wording sound technical?’ but ‘Does this wording still describe the same kind of evidence under the same conditions?’ Another useful diagnostic is a source-to-target trace: highlight every value and qualifier in the source, then annotate the exact passage that carries it in the target. An unpaired item is a question, not a cosmetic omission.

How meaning fails. Merging all stages into a single present-tense narrative destroys causality. Consider what a reader could reasonably infer from that mistake: the actor, the point in the lifecycle, or the numerical interpretation may shift without any obvious grammatical error. The remedy is to preserve the evidence-bearing distinction before editing for style. Construct a dated timeline that retains each event’s actor and evidence. A translator who cannot establish the source meaning should label the decision pending rather than select a plausible expression from general vocabulary.

Practical check. Review publication time against event time and subsequent corrections. Write a tiny audit receipt with the source field or phrase, its approved target expression, the source reference, the reviewer and the release status. Read the target alone and ask whether it implies a new action, a new certainty, a new unit or an unsupported comparison. This is the distinction between fidelity and fluency. Both are desirable, but fluent wording is not a substitute for a traceable scientific claim.

59. Reentry versus close approach

What it means. Atmospheric reentry risk and on-orbit conjunction risk are different assessment domains. A translated report should establish the object type and source authority before introducing a friendly synonym. The operational question is not ‘Does this wording sound technical?’ but ‘Does this wording still describe the same kind of evidence under the same conditions?’ The risk is especially high in mixed documents that contain an immutable machine layer and a human commentary layer. The correct treatment is not to translate everything identically: preserve parser-level tokens while explaining human meaning.

How meaning fails. A news summary about ‘debris danger’ may fuse a close-approach event with ground casualty risk. Consider what a reader could reasonably infer from that mistake: the actor, the point in the lifecycle, or the numerical interpretation may shift without any obvious grammatical error. The remedy is to preserve the evidence-bearing distinction before editing for style. Name the assessed hazard and avoid borrowing probabilities across contexts. A translator who cannot establish the source meaning should label the decision pending rather than select a plausible expression from general vocabulary.

Practical check. Route readers to separate assessments for collision and reentry. Write a tiny audit receipt with the source field or phrase, its approved target expression, the source reference, the reviewer and the release status. Read the target alone and ask whether it implies a new action, a new certainty, a new unit or an unsupported comparison. The intended audience may use different terms, but it must not receive a different event. Explain the concept at the surface while holding its data definition steady.

60. Research versus operations

What it means. Scientific literature explains models, whereas active operators use current validated mission procedures. A translated report should establish the object type and source authority before introducing a friendly synonym. The operational question is not ‘Does this wording sound technical?’ but ‘Does this wording still describe the same kind of evidence under the same conditions?’ A strong explanation for a general reader should move from observation to prediction to possible decision, explicitly naming the step being discussed. That sequence helps readers understand why a forecast is not a command.

How meaning fails. A tutorial calculation copied into a live alert can be mistaken for authorized operational guidance. Consider what a reader could reasonably infer from that mistake: the actor, the point in the lifecycle, or the numerical interpretation may shift without any obvious grammatical error. The remedy is to preserve the evidence-bearing distinction before editing for style. Label educational figures fictional and refer actual decisions to responsible teams. A translator who cannot establish the source meaning should label the decision pending rather than select a plausible expression from general vocabulary.

Practical check. Enforce a no-live-maneuver-instruction publishing gate. Write a tiny audit receipt with the source field or phrase, its approved target expression, the source reference, the reviewer and the release status. Read the target alone and ask whether it implies a new action, a new certainty, a new unit or an unsupported comparison. A real operational desk needs a controlled source, not a memory of an earlier alert. The audit should preserve enough provenance to recover what was known at each decision point.

61. Correction and withdrawal

What it means. An inaccurate translated alert may persist after the source is corrected. A translated report should establish the object type and source authority before introducing a friendly synonym. The operational question is not ‘Does this wording sound technical?’ but ‘Does this wording still describe the same kind of evidence under the same conditions?’ When two values are numerically plausible, a copied typo may escape ordinary proofreading. A reviewer should therefore compare identifiers, signs, time references and units as independent evidence, not as part of flowing prose.

How meaning fails. Quietly editing the headline without an audit record leaves outdated copies untraceable. Consider what a reader could reasonably infer from that mistake: the actor, the point in the lifecycle, or the numerical interpretation may shift without any obvious grammatical error. The remedy is to preserve the evidence-bearing distinction before editing for style. Issue a dated correction with superseded versions and replacement wording. A translator who cannot establish the source meaning should label the decision pending rather than select a plausible expression from general vocabulary.

Practical check. Verify all channels point to the new authoritative version. Write a tiny audit receipt with the source field or phrase, its approved target expression, the source reference, the reviewer and the release status. Read the target alone and ask whether it implies a new action, a new certainty, a new unit or an unsupported comparison. Do not approve this field on vocabulary similarity alone. Mark it verified only when the source-to-target evidence trail supports the same assertion and the same limits.

62. Release ownership

What it means. A technical translation passes through different specialists before external release. A translated report should establish the object type and source authority before introducing a friendly synonym. The operational question is not ‘Does this wording sound technical?’ but ‘Does this wording still describe the same kind of evidence under the same conditions?’ Plain language can carry exact science if it does not hide the conditional structure. Shorten long clauses, but keep the phrases that specify model, event, likelihood, revision and authority in the final sentence.

How meaning fails. A linguist alone cannot validate orbital dynamics; an engineer alone cannot ensure target-language clarity. Consider what a reader could reasonably infer from that mistake: the actor, the point in the lifecycle, or the numerical interpretation may shift without any obvious grammatical error. The remedy is to preserve the evidence-bearing distinction before editing for style. Require complementary linguistic, technical and publication reviews. A translator who cannot establish the source meaning should label the decision pending rather than select a plausible expression from general vocabulary.

Practical check. Collect signatures or explicit review receipts for critical claims before release. Write a tiny audit receipt with the source field or phrase, its approved target expression, the source reference, the reviewer and the release status. Read the target alone and ask whether it implies a new action, a new certainty, a new unit or an unsupported comparison. The release test is deliberately narrow: can another competent reviewer recover the same field, relationship, time and state without guessing? If not, return the item for clarification.

The repeatable method: twenty-eight translation and quality gates

Use these gates in order when the stakes are high. A low-risk historical explanation may combine some reviews; a time-sensitive alert should not bypass the checks that protect source identity, timing, uncertainty, authority or release. These are editorial gates, not operational flight rules, and the operator’s own safety system remains definitive.

1. Define the document class

Action: Record whether the input is a structured CCSDS data message, orbit-element delivery, human analyst note, public warning, educational explainer or retrospective incident review. A document label determines which strings are machine tokens, which sentences are prose and which authority can legitimately speak in the first person. In a practical production ledger, record an owner for this gate, the exact artifact inspected, the answer obtained and whether an unresolved question prevents release. The value of a gate lies in a checkable outcome; writing ‘done’ without evidence adds no safety.

A good reviewer can reproduce the result from the source and target, not from the translator’s memory. Where the evidence is missing, the outcome is hold and query, not a guessed pass. The next stage should consume only the approved version and retain the link to the previous stage, so a later change in an orbit solution or event status does not get hidden by an otherwise fluent paragraph.

2. Record the authorized source

Action: Save the provider, URL or file reference, retrieval timestamp and content checksum where available. A translator must be able to show exactly which snapshot of an evolving conjunction assessment was reviewed. In a practical production ledger, record an owner for this gate, the exact artifact inspected, the answer obtained and whether an unresolved question prevents release. The value of a gate lies in a checkable outcome; writing ‘done’ without evidence adds no safety.

A good reviewer can reproduce the result from the source and target, not from the translator’s memory. Where the evidence is missing, the outcome is hold and query, not a guessed pass. The next stage should consume only the approved version and retain the link to the previous stage, so a later change in an orbit solution or event status does not get hidden by an otherwise fluent paragraph.

3. Capture the version

Action: Identify message version, creation timestamp, event or pair key, and whether a later correction exists. A body that reads well but describes a superseded CDM is not release-ready. In a practical production ledger, record an owner for this gate, the exact artifact inspected, the answer obtained and whether an unresolved question prevents release. The value of a gate lies in a checkable outcome; writing ‘done’ without evidence adds no safety.

A good reviewer can reproduce the result from the source and target, not from the translator’s memory. Where the evidence is missing, the outcome is hold and query, not a guessed pass. The next stage should consume only the approved version and retain the link to the previous stage, so a later change in an orbit solution or event status does not get hidden by an otherwise fluent paragraph.

4. Parse the two objects

Action: Map primary and secondary IDs, type labels and source-specific aliases before translating narrative text. An object reversal can make every later sentence about relative motion misleading. In a practical production ledger, record an owner for this gate, the exact artifact inspected, the answer obtained and whether an unresolved question prevents release. The value of a gate lies in a checkable outcome; writing ‘done’ without evidence adds no safety.

A good reviewer can reproduce the result from the source and target, not from the translator’s memory. Where the evidence is missing, the outcome is hold and query, not a guessed pass. The next stage should consume only the approved version and retain the link to the previous stage, so a later change in an orbit solution or event status does not get hidden by an otherwise fluent paragraph.

5. Build a time ledger

Action: Separate observation span, orbit epoch, screening interval, CDM creation time, TCA, review time and publication time. The word ‘today’ is unsafe unless its clock and audience are explicitly determined. In a practical production ledger, record an owner for this gate, the exact artifact inspected, the answer obtained and whether an unresolved question prevents release. The value of a gate lies in a checkable outcome; writing ‘done’ without evidence adds no safety.

A good reviewer can reproduce the result from the source and target, not from the translator’s memory. Where the evidence is missing, the outcome is hold and query, not a guessed pass. The next stage should consume only the approved version and retain the link to the previous stage, so a later change in an orbit solution or event status does not get hidden by an otherwise fluent paragraph.

6. Preserve time scale

Action: Keep UTC or another declared time system in machine fields, and mark any reader-facing conversion. Even a perfect time conversion can mislead if it lacks a timezone label. In a practical production ledger, record an owner for this gate, the exact artifact inspected, the answer obtained and whether an unresolved question prevents release. The value of a gate lies in a checkable outcome; writing ‘done’ without evidence adds no safety.

A good reviewer can reproduce the result from the source and target, not from the translator’s memory. Where the evidence is missing, the outcome is hold and query, not a guessed pass. The next stage should consume only the approved version and retain the link to the previous stage, so a later change in an orbit solution or event status does not get hidden by an otherwise fluent paragraph.

7. Preserve the units

Action: List distance, velocity, covariance, area and angular units and prohibit implicit conversions. A numeric value without its dimension is only typography, not a safe technical statement. In a practical production ledger, record an owner for this gate, the exact artifact inspected, the answer obtained and whether an unresolved question prevents release. The value of a gate lies in a checkable outcome; writing ‘done’ without evidence adds no safety.

A good reviewer can reproduce the result from the source and target, not from the translator’s memory. Where the evidence is missing, the outcome is hold and query, not a guessed pass. The next stage should consume only the approved version and retain the link to the previous stage, so a later change in an orbit solution or event status does not get hidden by an otherwise fluent paragraph.

8. Identify reference frame

Action: Map the frame and axis conventions of every vector and matrix. A sign and number can be copied faithfully but assigned to the wrong physical direction. In a practical production ledger, record an owner for this gate, the exact artifact inspected, the answer obtained and whether an unresolved question prevents release. The value of a gate lies in a checkable outcome; writing ‘done’ without evidence adds no safety.

A good reviewer can reproduce the result from the source and target, not from the translator’s memory. Where the evidence is missing, the outcome is hold and query, not a guessed pass. The next stage should consume only the approved version and retain the link to the previous stage, so a later change in an orbit solution or event status does not get hidden by an otherwise fluent paragraph.

9. Distinguish observed from predicted

Action: Label measurements, orbit estimates, propagated states, screening outputs and analyst judgments separately. Readers should be able to tell what was directly recorded and what depends on a forecast. In a practical production ledger, record an owner for this gate, the exact artifact inspected, the answer obtained and whether an unresolved question prevents release. The value of a gate lies in a checkable outcome; writing ‘done’ without evidence adds no safety.

A good reviewer can reproduce the result from the source and target, not from the translator’s memory. Where the evidence is missing, the outcome is hold and query, not a guessed pass. The next stage should consume only the approved version and retain the link to the previous stage, so a later change in an orbit solution or event status does not get hidden by an otherwise fluent paragraph.

10. Copy raw identifiers

Action: Keep catalog numbers, international designators, field keys and data-package identifiers unchanged. Human display labels may be localized, but the identity join must survive every edition. In a practical production ledger, record an owner for this gate, the exact artifact inspected, the answer obtained and whether an unresolved question prevents release. The value of a gate lies in a checkable outcome; writing ‘done’ without evidence adds no safety.

A good reviewer can reproduce the result from the source and target, not from the translator’s memory. Where the evidence is missing, the outcome is hold and query, not a guessed pass. The next stage should consume only the approved version and retain the link to the previous stage, so a later change in an orbit solution or event status does not get hidden by an otherwise fluent paragraph.

11. Validate modern catalog coverage

Action: Check whether a legacy TLE interface omits six-digit objects, and prefer a supported alternative when the provider requires it. The absence of a record from a limited representation must not be narrated as the absence of an object. In a practical production ledger, record an owner for this gate, the exact artifact inspected, the answer obtained and whether an unresolved question prevents release. The value of a gate lies in a checkable outcome; writing ‘done’ without evidence adds no safety.

A good reviewer can reproduce the result from the source and target, not from the translator’s memory. Where the evidence is missing, the outcome is hold and query, not a guessed pass. The next stage should consume only the approved version and retain the link to the previous stage, so a later change in an orbit solution or event status does not get hidden by an otherwise fluent paragraph.

12. Separate CDM from OMM

Action: Treat a conjunction message and an orbit element message as different products serving different purposes. Replacing one with a familiar name for the other silently changes what the file can prove. In a practical production ledger, record an owner for this gate, the exact artifact inspected, the answer obtained and whether an unresolved question prevents release. The value of a gate lies in a checkable outcome; writing ‘done’ without evidence adds no safety.

A good reviewer can reproduce the result from the source and target, not from the translator’s memory. Where the evidence is missing, the outcome is hold and query, not a guessed pass. The next stage should consume only the approved version and retain the link to the previous stage, so a later change in an orbit solution or event status does not get hidden by an otherwise fluent paragraph.

13. Extract the event geometry

Action: Map TCA, predicted miss distance, relative position, relative speed and relevant units. Do not calculate risk or issue new flight recommendations from the translation layer. In a practical production ledger, record an owner for this gate, the exact artifact inspected, the answer obtained and whether an unresolved question prevents release. The value of a gate lies in a checkable outcome; writing ‘done’ without evidence adds no safety.

A good reviewer can reproduce the result from the source and target, not from the translator’s memory. Where the evidence is missing, the outcome is hold and query, not a guessed pass. The next stage should consume only the approved version and retain the link to the previous stage, so a later change in an orbit solution or event status does not get hidden by an otherwise fluent paragraph.

14. Explain covariance

Action: Write a reader-friendly explanation while retaining the matrix, reference frame and uncertainty status. Never reduce a multidimensional uncertainty structure to a single invented margin of error. In a practical production ledger, record an owner for this gate, the exact artifact inspected, the answer obtained and whether an unresolved question prevents release. The value of a gate lies in a checkable outcome; writing ‘done’ without evidence adds no safety.

A good reviewer can reproduce the result from the source and target, not from the translator’s memory. Where the evidence is missing, the outcome is hold and query, not a guessed pass. The next stage should consume only the approved version and retain the link to the previous stage, so a later change in an orbit solution or event status does not get hidden by an otherwise fluent paragraph.

15. State probability carefully

Action: Preserve the numerical probability and its meaning as a modeled estimate for that encounter. Avoid overconfident verbs such as ‘will hit’, ‘cannot hit’ or ‘is safe’ unless the cited authority states them within their scope. In a practical production ledger, record an owner for this gate, the exact artifact inspected, the answer obtained and whether an unresolved question prevents release. The value of a gate lies in a checkable outcome; writing ‘done’ without evidence adds no safety.

A good reviewer can reproduce the result from the source and target, not from the translator’s memory. Where the evidence is missing, the outcome is hold and query, not a guessed pass. The next stage should consume only the approved version and retain the link to the previous stage, so a later change in an orbit solution or event status does not get hidden by an otherwise fluent paragraph.

16. Track assumptions

Action: Capture model version, hard-body radius, missing parameters and data quality caveats. Risk language without assumption context may overstate comparability across sources. In a practical production ledger, record an owner for this gate, the exact artifact inspected, the answer obtained and whether an unresolved question prevents release. The value of a gate lies in a checkable outcome; writing ‘done’ without evidence adds no safety.

A good reviewer can reproduce the result from the source and target, not from the translator’s memory. Where the evidence is missing, the outcome is hold and query, not a guessed pass. The next stage should consume only the approved version and retain the link to the previous stage, so a later change in an orbit solution or event status does not get hidden by an otherwise fluent paragraph.

17. Lock field terminology

Action: Create a controlled list of source terms, definitions, target equivalents and forbidden false friends. A termbase should explain the concept, not only present a bilingual replacement word. In a practical production ledger, record an owner for this gate, the exact artifact inspected, the answer obtained and whether an unresolved question prevents release. The value of a gate lies in a checkable outcome; writing ‘done’ without evidence adds no safety.

A good reviewer can reproduce the result from the source and target, not from the translator’s memory. Where the evidence is missing, the outcome is hold and query, not a guessed pass. The next stage should consume only the approved version and retain the link to the previous stage, so a later change in an orbit solution or event status does not get hidden by an otherwise fluent paragraph.

18. Draft human prose

Action: Explain the event in clear sentences that retain status, time, evidence and uncertainty. Aim for accurate first reading; clarity is not achieved by deleting the condition that makes the claim true. In a practical production ledger, record an owner for this gate, the exact artifact inspected, the answer obtained and whether an unresolved question prevents release. The value of a gate lies in a checkable outcome; writing ‘done’ without evidence adds no safety.

A good reviewer can reproduce the result from the source and target, not from the translator’s memory. Where the evidence is missing, the outcome is hold and query, not a guessed pass. The next stage should consume only the approved version and retain the link to the previous stage, so a later change in an orbit solution or event status does not get hidden by an otherwise fluent paragraph.

19. Audit numerical invariants

Action: Compare signs, magnitudes, exponents, units, IDs, dates and versions independently. Numeric QA must be more exacting than a general spelling pass because errors can look syntactically smooth. In a practical production ledger, record an owner for this gate, the exact artifact inspected, the answer obtained and whether an unresolved question prevents release. The value of a gate lies in a checkable outcome; writing ‘done’ without evidence adds no safety.

A good reviewer can reproduce the result from the source and target, not from the translator’s memory. Where the evidence is missing, the outcome is hold and query, not a guessed pass. The next stage should consume only the approved version and retain the link to the previous stage, so a later change in an orbit solution or event status does not get hidden by an otherwise fluent paragraph.

20. Audit epistemic verbs

Action: Check could, may, predicted, detected, assessed, proposed, approved, commanded and executed. Every change of verb can move the record into a different operational stage. In a practical production ledger, record an owner for this gate, the exact artifact inspected, the answer obtained and whether an unresolved question prevents release. The value of a gate lies in a checkable outcome; writing ‘done’ without evidence adds no safety.

A good reviewer can reproduce the result from the source and target, not from the translator’s memory. Where the evidence is missing, the outcome is hold and query, not a guessed pass. The next stage should consume only the approved version and retain the link to the previous stage, so a later change in an orbit solution or event status does not get hidden by an otherwise fluent paragraph.

21. Perform domain review

Action: Have a qualified space-safety or flight-dynamics reviewer check the factual interpretation. The translator should not infer that a bilingual review alone validates orbital dynamics. In a practical production ledger, record an owner for this gate, the exact artifact inspected, the answer obtained and whether an unresolved question prevents release. The value of a gate lies in a checkable outcome; writing ‘done’ without evidence adds no safety.

A good reviewer can reproduce the result from the source and target, not from the translator’s memory. Where the evidence is missing, the outcome is hold and query, not a guessed pass. The next stage should consume only the approved version and retain the link to the previous stage, so a later change in an orbit solution or event status does not get hidden by an otherwise fluent paragraph.

22. Perform independent language review

Action: Have a competent target-language reviewer examine ambiguity, register, accessibility and readability. An engineer may know a field perfectly but miss an unintended implication in the target language. In a practical production ledger, record an owner for this gate, the exact artifact inspected, the answer obtained and whether an unresolved question prevents release. The value of a gate lies in a checkable outcome; writing ‘done’ without evidence adds no safety.

A good reviewer can reproduce the result from the source and target, not from the translator’s memory. Where the evidence is missing, the outcome is hold and query, not a guessed pass. The next stage should consume only the approved version and retain the link to the previous stage, so a later change in an orbit solution or event status does not get hidden by an otherwise fluent paragraph.

23. Simulate a non-specialist reading

Action: Ask a reader to explain what the message says happened, is predicted and remains unknown. A response that describes a confirmed crash when the source reports a predicted approach fails this gate. In a practical production ledger, record an owner for this gate, the exact artifact inspected, the answer obtained and whether an unresolved question prevents release. The value of a gate lies in a checkable outcome; writing ‘done’ without evidence adds no safety.

A good reviewer can reproduce the result from the source and target, not from the translator’s memory. Where the evidence is missing, the outcome is hold and query, not a guessed pass. The next stage should consume only the approved version and retain the link to the previous stage, so a later change in an orbit solution or event status does not get hidden by an otherwise fluent paragraph.

24. Test rendering

Action: Inspect headings, tables, right-to-left passages, scientific notation, screen readers and copied values. A technically correct string may still be displayed ambiguously in a localized interface. In a practical production ledger, record an owner for this gate, the exact artifact inspected, the answer obtained and whether an unresolved question prevents release. The value of a gate lies in a checkable outcome; writing ‘done’ without evidence adds no safety.

A good reviewer can reproduce the result from the source and target, not from the translator’s memory. Where the evidence is missing, the outcome is hold and query, not a guessed pass. The next stage should consume only the approved version and retain the link to the previous stage, so a later change in an orbit solution or event status does not get hidden by an otherwise fluent paragraph.

25. Check permissions and audience

Action: Confirm whether detailed values are approved for public circulation and who authorized the release. Provenance and publication approval are separate from translation accuracy. In a practical production ledger, record an owner for this gate, the exact artifact inspected, the answer obtained and whether an unresolved question prevents release. The value of a gate lies in a checkable outcome; writing ‘done’ without evidence adds no safety.

A good reviewer can reproduce the result from the source and target, not from the translator’s memory. Where the evidence is missing, the outcome is hold and query, not a guessed pass. The next stage should consume only the approved version and retain the link to the previous stage, so a later change in an orbit solution or event status does not get hidden by an otherwise fluent paragraph.

26. Publish a versioned record

Action: Include revision date, source class, caveats and clear links to authoritative information. A public page should not masquerade as a live alert feed or official decision system. In a practical production ledger, record an owner for this gate, the exact artifact inspected, the answer obtained and whether an unresolved question prevents release. The value of a gate lies in a checkable outcome; writing ‘done’ without evidence adds no safety.

A good reviewer can reproduce the result from the source and target, not from the translator’s memory. Where the evidence is missing, the outcome is hold and query, not a guessed pass. The next stage should consume only the approved version and retain the link to the previous stage, so a later change in an orbit solution or event status does not get hidden by an otherwise fluent paragraph.

27. Monitor for correction

Action: Watch for revised source data and define what triggers a translated correction or withdrawal. Publication is the start of an evidence lifecycle, not the end of responsibility. In a practical production ledger, record an owner for this gate, the exact artifact inspected, the answer obtained and whether an unresolved question prevents release. The value of a gate lies in a checkable outcome; writing ‘done’ without evidence adds no safety.

A good reviewer can reproduce the result from the source and target, not from the translator’s memory. Where the evidence is missing, the outcome is hold and query, not a guessed pass. The next stage should consume only the approved version and retain the link to the previous stage, so a later change in an orbit solution or event status does not get hidden by an otherwise fluent paragraph.

28. Archive the receipt

Action: Store source snapshot, target snapshot, termbase edition, numeric comparison and sign-off. Later reviewers should be able to explain why a sentence was defensible when released. In a practical production ledger, record an owner for this gate, the exact artifact inspected, the answer obtained and whether an unresolved question prevents release. The value of a gate lies in a checkable outcome; writing ‘done’ without evidence adds no safety.

A good reviewer can reproduce the result from the source and target, not from the translator’s memory. Where the evidence is missing, the outcome is hold and query, not a guessed pass. The next stage should consume only the approved version and retain the link to the previous stage, so a later change in an orbit solution or event status does not get hidden by an otherwise fluent paragraph.

Failure modes: fifty ways a fluent translation can change the event

A failure mode becomes useful when the reviewer can identify the mistaken inference, not merely circle the wrong word. These entries support a diagnostic checklist. The examples do not predict how frequently any error occurs; they identify foreseeable categories and corresponding controls.

1. Conjunction becomes collision

Symptom: A predicted pairing is called a confirmed impact. Why it matters: Misstates the basic event type. Repair: Replace with ‘predicted close approach’ and record the forecast basis. The safe review procedure is to compare the exact source assertion with the target assertion, identify which invariant was lost, and confirm that the replacement wording restores it. Do not treat this as a spelling-only fix: it belongs in the evidence and correction log so another reviewer can see why a version changed.

For this case, conduct a three-part test. First, read the source without the target and write what is established, what is estimated and what remains unknown. Second, read only the target and ask whether a technically trained reader would reach the same conclusion. Third, compare the data-bearing elements—especially identity, time, quantity, status and authorization. If a real document does not contain enough information to close the case, state the missing field and return the question to the source owner.

2. ‘May’ becomes ‘will’

Symptom: An uncertain event is rewritten in the future indicative. Why it matters: Converts possibility into certainty. Repair: Restore modal force and source conditions. The safe review procedure is to compare the exact source assertion with the target assertion, identify which invariant was lost, and confirm that the replacement wording restores it. Do not treat this as a spelling-only fix: it belongs in the evidence and correction log so another reviewer can see why a version changed.

For this case, conduct a three-part test. First, read the source without the target and write what is established, what is estimated and what remains unknown. Second, read only the target and ask whether a technically trained reader would reach the same conclusion. Third, compare the data-bearing elements—especially identity, time, quantity, status and authorization. If a real document does not contain enough information to close the case, state the missing field and return the question to the source owner.

3. Probability becomes guarantee

Symptom: A small modeled probability is phrased as impossible. Why it matters: Overstates confidence. Repair: Use conditional model language with provenance. The safe review procedure is to compare the exact source assertion with the target assertion, identify which invariant was lost, and confirm that the replacement wording restores it. Do not treat this as a spelling-only fix: it belongs in the evidence and correction log so another reviewer can see why a version changed.

For this case, conduct a three-part test. First, read the source without the target and write what is established, what is estimated and what remains unknown. Second, read only the target and ask whether a technically trained reader would reach the same conclusion. Third, compare the data-bearing elements—especially identity, time, quantity, status and authorization. If a real document does not contain enough information to close the case, state the missing field and return the question to the source owner.

4. Probability exponent lost

Symptom: 2E-5 is printed as 2E5. Why it matters: Changes magnitude catastrophically. Repair: Compare scientific notation character by character. The safe review procedure is to compare the exact source assertion with the target assertion, identify which invariant was lost, and confirm that the replacement wording restores it. Do not treat this as a spelling-only fix: it belongs in the evidence and correction log so another reviewer can see why a version changed.

For this case, conduct a three-part test. First, read the source without the target and write what is established, what is estimated and what remains unknown. Second, read only the target and ask whether a technically trained reader would reach the same conclusion. Third, compare the data-bearing elements—especially identity, time, quantity, status and authorization. If a real document does not contain enough information to close the case, state the missing field and return the question to the source owner.

5. Percent conversion wrong

Symptom: 0.0001 is presented as 0.0001 percent. Why it matters: Creates a hundredfold interpretive difference. Repair: State raw decimal and verify any percent rendering. The safe review procedure is to compare the exact source assertion with the target assertion, identify which invariant was lost, and confirm that the replacement wording restores it. Do not treat this as a spelling-only fix: it belongs in the evidence and correction log so another reviewer can see why a version changed.

For this case, conduct a three-part test. First, read the source without the target and write what is established, what is estimated and what remains unknown. Second, read only the target and ask whether a technically trained reader would reach the same conclusion. Third, compare the data-bearing elements—especially identity, time, quantity, status and authorization. If a real document does not contain enough information to close the case, state the missing field and return the question to the source owner.

6. Covariance treated as standard deviation

Symptom: Variance entry is read as a linear one-sigma error. Why it matters: Breaks dimensional interpretation. Repair: Retain covariance definition and validate any derived numbers. The safe review procedure is to compare the exact source assertion with the target assertion, identify which invariant was lost, and confirm that the replacement wording restores it. Do not treat this as a spelling-only fix: it belongs in the evidence and correction log so another reviewer can see why a version changed.

For this case, conduct a three-part test. First, read the source without the target and write what is established, what is estimated and what remains unknown. Second, read only the target and ask whether a technically trained reader would reach the same conclusion. Third, compare the data-bearing elements—especially identity, time, quantity, status and authorization. If a real document does not contain enough information to close the case, state the missing field and return the question to the source owner.

7. Off-diagonal entries removed

Symptom: Only the variance diagonal survives translation. Why it matters: Discards correlation information. Repair: Preserve complete matrix or identify intentional omission. The safe review procedure is to compare the exact source assertion with the target assertion, identify which invariant was lost, and confirm that the replacement wording restores it. Do not treat this as a spelling-only fix: it belongs in the evidence and correction log so another reviewer can see why a version changed.

For this case, conduct a three-part test. First, read the source without the target and write what is established, what is estimated and what remains unknown. Second, read only the target and ask whether a technically trained reader would reach the same conclusion. Third, compare the data-bearing elements—especially identity, time, quantity, status and authorization. If a real document does not contain enough information to close the case, state the missing field and return the question to the source owner.

8. Frame omitted

Symptom: RTN data are explained as global Cartesian axes. Why it matters: Reassigns physical meaning to components. Repair: Keep reference-frame identification beside the values. The safe review procedure is to compare the exact source assertion with the target assertion, identify which invariant was lost, and confirm that the replacement wording restores it. Do not treat this as a spelling-only fix: it belongs in the evidence and correction log so another reviewer can see why a version changed.

For this case, conduct a three-part test. First, read the source without the target and write what is established, what is estimated and what remains unknown. Second, read only the target and ask whether a technically trained reader would reach the same conclusion. Third, compare the data-bearing elements—especially identity, time, quantity, status and authorization. If a real document does not contain enough information to close the case, state the missing field and return the question to the source owner.

9. Sign reversed

Symptom: A negative component is copied without its sign. Why it matters: Changes relative geometry. Repair: Perform sign-level comparison. The safe review procedure is to compare the exact source assertion with the target assertion, identify which invariant was lost, and confirm that the replacement wording restores it. Do not treat this as a spelling-only fix: it belongs in the evidence and correction log so another reviewer can see why a version changed.

For this case, conduct a three-part test. First, read the source without the target and write what is established, what is estimated and what remains unknown. Second, read only the target and ask whether a technically trained reader would reach the same conclusion. Third, compare the data-bearing elements—especially identity, time, quantity, status and authorization. If a real document does not contain enough information to close the case, state the missing field and return the question to the source owner.

10. Primary and secondary swapped

Symptom: Object labels are reversed in the prose. Why it matters: Moves risk attribution to the wrong object. Repair: Check object block mapping before publication. The safe review procedure is to compare the exact source assertion with the target assertion, identify which invariant was lost, and confirm that the replacement wording restores it. Do not treat this as a spelling-only fix: it belongs in the evidence and correction log so another reviewer can see why a version changed.

For this case, conduct a three-part test. First, read the source without the target and write what is established, what is estimated and what remains unknown. Second, read only the target and ask whether a technically trained reader would reach the same conclusion. Third, compare the data-bearing elements—especially identity, time, quantity, status and authorization. If a real document does not contain enough information to close the case, state the missing field and return the question to the source owner.

11. Message date used as TCA

Symptom: Creation timestamp is described as encounter time. Why it matters: Moves the event to a different instant. Repair: Maintain separate timestamp ledger. The safe review procedure is to compare the exact source assertion with the target assertion, identify which invariant was lost, and confirm that the replacement wording restores it. Do not treat this as a spelling-only fix: it belongs in the evidence and correction log so another reviewer can see why a version changed.

For this case, conduct a three-part test. First, read the source without the target and write what is established, what is estimated and what remains unknown. Second, read only the target and ask whether a technically trained reader would reach the same conclusion. Third, compare the data-bearing elements—especially identity, time, quantity, status and authorization. If a real document does not contain enough information to close the case, state the missing field and return the question to the source owner.

12. Local time unlabelled

Symptom: A UTC display is converted with no timezone. Why it matters: Invites wrong date or hour interpretation. Repair: Preserve UTC and label conversion explicitly. The safe review procedure is to compare the exact source assertion with the target assertion, identify which invariant was lost, and confirm that the replacement wording restores it. Do not treat this as a spelling-only fix: it belongs in the evidence and correction log so another reviewer can see why a version changed.

For this case, conduct a three-part test. First, read the source without the target and write what is established, what is estimated and what remains unknown. Second, read only the target and ask whether a technically trained reader would reach the same conclusion. Third, compare the data-bearing elements—especially identity, time, quantity, status and authorization. If a real document does not contain enough information to close the case, state the missing field and return the question to the source owner.

13. Fractional seconds dropped

Symptom: Precision is silently reduced in critical fields. Why it matters: Masks possible ordering differences. Repair: Keep raw machine precision. The safe review procedure is to compare the exact source assertion with the target assertion, identify which invariant was lost, and confirm that the replacement wording restores it. Do not treat this as a spelling-only fix: it belongs in the evidence and correction log so another reviewer can see why a version changed.

For this case, conduct a three-part test. First, read the source without the target and write what is established, what is estimated and what remains unknown. Second, read only the target and ask whether a technically trained reader would reach the same conclusion. Third, compare the data-bearing elements—especially identity, time, quantity, status and authorization. If a real document does not contain enough information to close the case, state the missing field and return the question to the source owner.

14. Screening interval becomes maneuver window

Symptom: Analysis bounds are narrated as a recommended burn period. Why it matters: Invents operational instruction. Repair: Label screening purpose correctly. The safe review procedure is to compare the exact source assertion with the target assertion, identify which invariant was lost, and confirm that the replacement wording restores it. Do not treat this as a spelling-only fix: it belongs in the evidence and correction log so another reviewer can see why a version changed.

For this case, conduct a three-part test. First, read the source without the target and write what is established, what is estimated and what remains unknown. Second, read only the target and ask whether a technically trained reader would reach the same conclusion. Third, compare the data-bearing elements—especially identity, time, quantity, status and authorization. If a real document does not contain enough information to close the case, state the missing field and return the question to the source owner.

15. Miss distance described as present separation

Symptom: Forecast minimum is called current physical distance. Why it matters: Changes time and epistemic status. Repair: Say predicted minimum separation. The safe review procedure is to compare the exact source assertion with the target assertion, identify which invariant was lost, and confirm that the replacement wording restores it. Do not treat this as a spelling-only fix: it belongs in the evidence and correction log so another reviewer can see why a version changed.

For this case, conduct a three-part test. First, read the source without the target and write what is established, what is estimated and what remains unknown. Second, read only the target and ask whether a technically trained reader would reach the same conclusion. Third, compare the data-bearing elements—especially identity, time, quantity, status and authorization. If a real document does not contain enough information to close the case, state the missing field and return the question to the source owner.

16. Relative speed becomes absolute speed

Symptom: Encounter-relative quantity is called spacecraft speed. Why it matters: Changes physical definition. Repair: Add ‘relative’ and preserve unit. The safe review procedure is to compare the exact source assertion with the target assertion, identify which invariant was lost, and confirm that the replacement wording restores it. Do not treat this as a spelling-only fix: it belongs in the evidence and correction log so another reviewer can see why a version changed.

For this case, conduct a three-part test. First, read the source without the target and write what is established, what is estimated and what remains unknown. Second, read only the target and ask whether a technically trained reader would reach the same conclusion. Third, compare the data-bearing elements—especially identity, time, quantity, status and authorization. If a real document does not contain enough information to close the case, state the missing field and return the question to the source owner.

17. Issuer mistaken for operator

Symptom: Provider’s alert is attributed to spacecraft control. Why it matters: Invents an authority relationship. Repair: Name originator and decision-maker separately. The safe review procedure is to compare the exact source assertion with the target assertion, identify which invariant was lost, and confirm that the replacement wording restores it. Do not treat this as a spelling-only fix: it belongs in the evidence and correction log so another reviewer can see why a version changed.

For this case, conduct a three-part test. First, read the source without the target and write what is established, what is estimated and what remains unknown. Second, read only the target and ask whether a technically trained reader would reach the same conclusion. Third, compare the data-bearing elements—especially identity, time, quantity, status and authorization. If a real document does not contain enough information to close the case, state the missing field and return the question to the source owner.

18. Recommendation becomes approval

Symptom: A proposal is announced as approved. Why it matters: Changes procedural status. Repair: Preserve conditional and sign-off language. The safe review procedure is to compare the exact source assertion with the target assertion, identify which invariant was lost, and confirm that the replacement wording restores it. Do not treat this as a spelling-only fix: it belongs in the evidence and correction log so another reviewer can see why a version changed.

For this case, conduct a three-part test. First, read the source without the target and write what is established, what is estimated and what remains unknown. Second, read only the target and ask whether a technically trained reader would reach the same conclusion. Third, compare the data-bearing elements—especially identity, time, quantity, status and authorization. If a real document does not contain enough information to close the case, state the missing field and return the question to the source owner.

19. Planned becomes executed

Symptom: A possible maneuver is written as a completed action. Why it matters: Creates false event history. Repair: Require execution evidence. The safe review procedure is to compare the exact source assertion with the target assertion, identify which invariant was lost, and confirm that the replacement wording restores it. Do not treat this as a spelling-only fix: it belongs in the evidence and correction log so another reviewer can see why a version changed.

For this case, conduct a three-part test. First, read the source without the target and write what is established, what is estimated and what remains unknown. Second, read only the target and ask whether a technically trained reader would reach the same conclusion. Third, compare the data-bearing elements—especially identity, time, quantity, status and authorization. If a real document does not contain enough information to close the case, state the missing field and return the question to the source owner.

20. Updated alert treated as duplicate

Symptom: New risk data are ignored because pair IDs are unchanged. Why it matters: Suppresses revision. Repair: Compare message IDs and generation times. The safe review procedure is to compare the exact source assertion with the target assertion, identify which invariant was lost, and confirm that the replacement wording restores it. Do not treat this as a spelling-only fix: it belongs in the evidence and correction log so another reviewer can see why a version changed.

For this case, conduct a three-part test. First, read the source without the target and write what is established, what is estimated and what remains unknown. Second, read only the target and ask whether a technically trained reader would reach the same conclusion. Third, compare the data-bearing elements—especially identity, time, quantity, status and authorization. If a real document does not contain enough information to close the case, state the missing field and return the question to the source owner.

21. Superseded alert republished

Symptom: Old warning appears as latest analysis. Why it matters: Misleads about currency. Repair: Bind public copy to source version. The safe review procedure is to compare the exact source assertion with the target assertion, identify which invariant was lost, and confirm that the replacement wording restores it. Do not treat this as a spelling-only fix: it belongs in the evidence and correction log so another reviewer can see why a version changed.

For this case, conduct a three-part test. First, read the source without the target and write what is established, what is estimated and what remains unknown. Second, read only the target and ask whether a technically trained reader would reach the same conclusion. Third, compare the data-bearing elements—especially identity, time, quantity, status and authorization. If a real document does not contain enough information to close the case, state the missing field and return the question to the source owner.

22. Blank treated as zero

Symptom: Missing probability is paraphrased as zero risk. Why it matters: Converts missing evidence into reassuring result. Repair: Represent not-reported explicitly. The safe review procedure is to compare the exact source assertion with the target assertion, identify which invariant was lost, and confirm that the replacement wording restores it. Do not treat this as a spelling-only fix: it belongs in the evidence and correction log so another reviewer can see why a version changed.

For this case, conduct a three-part test. First, read the source without the target and write what is established, what is estimated and what remains unknown. Second, read only the target and ask whether a technically trained reader would reach the same conclusion. Third, compare the data-bearing elements—especially identity, time, quantity, status and authorization. If a real document does not contain enough information to close the case, state the missing field and return the question to the source owner.

23. TLE gap called absence

Symptom: Six-digit object’s legacy row is missing. Why it matters: Suggests object no longer exists. Repair: Explain format coverage limit. The safe review procedure is to compare the exact source assertion with the target assertion, identify which invariant was lost, and confirm that the replacement wording restores it. Do not treat this as a spelling-only fix: it belongs in the evidence and correction log so another reviewer can see why a version changed.

For this case, conduct a three-part test. First, read the source without the target and write what is established, what is estimated and what remains unknown. Second, read only the target and ask whether a technically trained reader would reach the same conclusion. Third, compare the data-bearing elements—especially identity, time, quantity, status and authorization. If a real document does not contain enough information to close the case, state the missing field and return the question to the source owner.

24. Parser assumes TLE default

Symptom: CSV data arrive at a TLE parser. Why it matters: Creates silent import failure. Repair: Specify FORMAT and validate schema. The safe review procedure is to compare the exact source assertion with the target assertion, identify which invariant was lost, and confirm that the replacement wording restores it. Do not treat this as a spelling-only fix: it belongs in the evidence and correction log so another reviewer can see why a version changed.

For this case, conduct a three-part test. First, read the source without the target and write what is established, what is estimated and what remains unknown. Second, read only the target and ask whether a technically trained reader would reach the same conclusion. Third, compare the data-bearing elements—especially identity, time, quantity, status and authorization. If a real document does not contain enough information to close the case, state the missing field and return the question to the source owner.

25. Catalog ID truncated

Symptom: 100000 is stored as 00000 or rejected. Why it matters: Breaks object identity. Repair: Use a width-safe ID field. The safe review procedure is to compare the exact source assertion with the target assertion, identify which invariant was lost, and confirm that the replacement wording restores it. Do not treat this as a spelling-only fix: it belongs in the evidence and correction log so another reviewer can see why a version changed.

For this case, conduct a three-part test. First, read the source without the target and write what is established, what is estimated and what remains unknown. Second, read only the target and ask whether a technically trained reader would reach the same conclusion. Third, compare the data-bearing elements—especially identity, time, quantity, status and authorization. If a real document does not contain enough information to close the case, state the missing field and return the question to the source owner.

26. Object name used as key

Symptom: Localized display name serves as unique ID. Why it matters: Can join records to the wrong satellite. Repair: Use stable machine identifier. The safe review procedure is to compare the exact source assertion with the target assertion, identify which invariant was lost, and confirm that the replacement wording restores it. Do not treat this as a spelling-only fix: it belongs in the evidence and correction log so another reviewer can see why a version changed.

For this case, conduct a three-part test. First, read the source without the target and write what is established, what is estimated and what remains unknown. Second, read only the target and ask whether a technically trained reader would reach the same conclusion. Third, compare the data-bearing elements—especially identity, time, quantity, status and authorization. If a real document does not contain enough information to close the case, state the missing field and return the question to the source owner.

27. OMM called TLE

Symptom: New format is described as identical legacy rows. Why it matters: Conceals metadata differences. Repair: Explain representation and required fields. The safe review procedure is to compare the exact source assertion with the target assertion, identify which invariant was lost, and confirm that the replacement wording restores it. Do not treat this as a spelling-only fix: it belongs in the evidence and correction log so another reviewer can see why a version changed.

For this case, conduct a three-part test. First, read the source without the target and write what is established, what is estimated and what remains unknown. Second, read only the target and ask whether a technically trained reader would reach the same conclusion. Third, compare the data-bearing elements—especially identity, time, quantity, status and authorization. If a real document does not contain enough information to close the case, state the missing field and return the question to the source owner.

28. Epoch confused with retrieval date

Symptom: A downloaded file is called a fresh orbit estimate. Why it matters: Misrepresents data freshness. Repair: Track both times. The safe review procedure is to compare the exact source assertion with the target assertion, identify which invariant was lost, and confirm that the replacement wording restores it. Do not treat this as a spelling-only fix: it belongs in the evidence and correction log so another reviewer can see why a version changed.

For this case, conduct a three-part test. First, read the source without the target and write what is established, what is estimated and what remains unknown. Second, read only the target and ask whether a technically trained reader would reach the same conclusion. Third, compare the data-bearing elements—especially identity, time, quantity, status and authorization. If a real document does not contain enough information to close the case, state the missing field and return the question to the source owner.

29. Source forecasts spliced

Symptom: One provider’s TCA is paired with another’s probability. Why it matters: Creates a record nobody issued. Repair: Maintain source-specific rows. The safe review procedure is to compare the exact source assertion with the target assertion, identify which invariant was lost, and confirm that the replacement wording restores it. Do not treat this as a spelling-only fix: it belongs in the evidence and correction log so another reviewer can see why a version changed.

For this case, conduct a three-part test. First, read the source without the target and write what is established, what is estimated and what remains unknown. Second, read only the target and ask whether a technically trained reader would reach the same conclusion. Third, compare the data-bearing elements—especially identity, time, quantity, status and authorization. If a real document does not contain enough information to close the case, state the missing field and return the question to the source owner.

30. Hard-body radius invented

Symptom: Public writer infers radius from imagery. Why it matters: Changes risk model assumption. Repair: Quote only source-supported model inputs. The safe review procedure is to compare the exact source assertion with the target assertion, identify which invariant was lost, and confirm that the replacement wording restores it. Do not treat this as a spelling-only fix: it belongs in the evidence and correction log so another reviewer can see why a version changed.

For this case, conduct a three-part test. First, read the source without the target and write what is established, what is estimated and what remains unknown. Second, read only the target and ask whether a technically trained reader would reach the same conclusion. Third, compare the data-bearing elements—especially identity, time, quantity, status and authorization. If a real document does not contain enough information to close the case, state the missing field and return the question to the source owner.

31. Color code decontextualized

Symptom: A local alert color is translated as universal. Why it matters: Changes risk policy semantics. Repair: Keep the issuing scheme label. The safe review procedure is to compare the exact source assertion with the target assertion, identify which invariant was lost, and confirm that the replacement wording restores it. Do not treat this as a spelling-only fix: it belongs in the evidence and correction log so another reviewer can see why a version changed.

For this case, conduct a three-part test. First, read the source without the target and write what is established, what is estimated and what remains unknown. Second, read only the target and ask whether a technically trained reader would reach the same conclusion. Third, compare the data-bearing elements—especially identity, time, quantity, status and authorization. If a real document does not contain enough information to close the case, state the missing field and return the question to the source owner.

32. Warning level inflated

Symptom: ‘Watch’ becomes ’emergency’ for excitement. Why it matters: Causes unwarranted alarm. Repair: Preserve exact local categorization. The safe review procedure is to compare the exact source assertion with the target assertion, identify which invariant was lost, and confirm that the replacement wording restores it. Do not treat this as a spelling-only fix: it belongs in the evidence and correction log so another reviewer can see why a version changed.

For this case, conduct a three-part test. First, read the source without the target and write what is established, what is estimated and what remains unknown. Second, read only the target and ask whether a technically trained reader would reach the same conclusion. Third, compare the data-bearing elements—especially identity, time, quantity, status and authorization. If a real document does not contain enough information to close the case, state the missing field and return the question to the source owner.

33. Warning level softened

Symptom: ‘Review required’ becomes ‘normal’. Why it matters: Can obscure source urgency. Repair: Maintain the actionable source distinction. The safe review procedure is to compare the exact source assertion with the target assertion, identify which invariant was lost, and confirm that the replacement wording restores it. Do not treat this as a spelling-only fix: it belongs in the evidence and correction log so another reviewer can see why a version changed.

For this case, conduct a three-part test. First, read the source without the target and write what is established, what is estimated and what remains unknown. Second, read only the target and ask whether a technically trained reader would reach the same conclusion. Third, compare the data-bearing elements—especially identity, time, quantity, status and authorization. If a real document does not contain enough information to close the case, state the missing field and return the question to the source owner.

34. Manual translation of schema keys

Symptom: OBJECT_ID is localized inside payload. Why it matters: Breaks machine interoperability. Repair: Keep raw keys and translate adjacent labels. The safe review procedure is to compare the exact source assertion with the target assertion, identify which invariant was lost, and confirm that the replacement wording restores it. Do not treat this as a spelling-only fix: it belongs in the evidence and correction log so another reviewer can see why a version changed.

For this case, conduct a three-part test. First, read the source without the target and write what is established, what is estimated and what remains unknown. Second, read only the target and ask whether a technically trained reader would reach the same conclusion. Third, compare the data-bearing elements—especially identity, time, quantity, status and authorization. If a real document does not contain enough information to close the case, state the missing field and return the question to the source owner.

35. Comments normalized into certainty

Symptom: Analyst shorthand ‘possible’ gets deleted. Why it matters: Loses caveat. Repair: Review prose against full context. The safe review procedure is to compare the exact source assertion with the target assertion, identify which invariant was lost, and confirm that the replacement wording restores it. Do not treat this as a spelling-only fix: it belongs in the evidence and correction log so another reviewer can see why a version changed.

For this case, conduct a three-part test. First, read the source without the target and write what is established, what is estimated and what remains unknown. Second, read only the target and ask whether a technically trained reader would reach the same conclusion. Third, compare the data-bearing elements—especially identity, time, quantity, status and authorization. If a real document does not contain enough information to close the case, state the missing field and return the question to the source owner.

36. Outdated translation memory

Symptom: Earlier TCA reused for later event. Why it matters: Contaminates current notice. Repair: Force variable and version audit. The safe review procedure is to compare the exact source assertion with the target assertion, identify which invariant was lost, and confirm that the replacement wording restores it. Do not treat this as a spelling-only fix: it belongs in the evidence and correction log so another reviewer can see why a version changed.

For this case, conduct a three-part test. First, read the source without the target and write what is established, what is estimated and what remains unknown. Second, read only the target and ask whether a technically trained reader would reach the same conclusion. Third, compare the data-bearing elements—especially identity, time, quantity, status and authorization. If a real document does not contain enough information to close the case, state the missing field and return the question to the source owner.

37. Machine translation drops negation

Symptom: ‘Not approved’ becomes ‘approved’. Why it matters: Inverts decision. Repair: Check every negation and status label. The safe review procedure is to compare the exact source assertion with the target assertion, identify which invariant was lost, and confirm that the replacement wording restores it. Do not treat this as a spelling-only fix: it belongs in the evidence and correction log so another reviewer can see why a version changed.

For this case, conduct a three-part test. First, read the source without the target and write what is established, what is estimated and what remains unknown. Second, read only the target and ask whether a technically trained reader would reach the same conclusion. Third, compare the data-bearing elements—especially identity, time, quantity, status and authorization. If a real document does not contain enough information to close the case, state the missing field and return the question to the source owner.

38. Decimal comma ambiguous

Symptom: 1,200 interpreted as 1.2 or 1200. Why it matters: Changes magnitude. Repair: Enforce locale-specific numeric rules. The safe review procedure is to compare the exact source assertion with the target assertion, identify which invariant was lost, and confirm that the replacement wording restores it. Do not treat this as a spelling-only fix: it belongs in the evidence and correction log so another reviewer can see why a version changed.

For this case, conduct a three-part test. First, read the source without the target and write what is established, what is estimated and what remains unknown. Second, read only the target and ask whether a technically trained reader would reach the same conclusion. Third, compare the data-bearing elements—especially identity, time, quantity, status and authorization. If a real document does not contain enough information to close the case, state the missing field and return the question to the source owner.

39. Bi-directional text reorder

Symptom: Mixed RTL and Latin identifiers display in odd order. Why it matters: Breaks copying and identity. Repair: Test rendered and underlying strings. The safe review procedure is to compare the exact source assertion with the target assertion, identify which invariant was lost, and confirm that the replacement wording restores it. Do not treat this as a spelling-only fix: it belongs in the evidence and correction log so another reviewer can see why a version changed.

For this case, conduct a three-part test. First, read the source without the target and write what is established, what is estimated and what remains unknown. Second, read only the target and ask whether a technically trained reader would reach the same conclusion. Third, compare the data-bearing elements—especially identity, time, quantity, status and authorization. If a real document does not contain enough information to close the case, state the missing field and return the question to the source owner.

40. Alt text asserts impact

Symptom: Graphic description says two satellites collided. Why it matters: Invents physical outcome. Repair: Describe predicted geometry and labels. The safe review procedure is to compare the exact source assertion with the target assertion, identify which invariant was lost, and confirm that the replacement wording restores it. Do not treat this as a spelling-only fix: it belongs in the evidence and correction log so another reviewer can see why a version changed.

For this case, conduct a three-part test. First, read the source without the target and write what is established, what is estimated and what remains unknown. Second, read only the target and ask whether a technically trained reader would reach the same conclusion. Third, compare the data-bearing elements—especially identity, time, quantity, status and authorization. If a real document does not contain enough information to close the case, state the missing field and return the question to the source owner.

41. Graph legend reversed

Symptom: Uncertainty bands switch between objects. Why it matters: Misattributes values. Repair: Audit legend against data table. The safe review procedure is to compare the exact source assertion with the target assertion, identify which invariant was lost, and confirm that the replacement wording restores it. Do not treat this as a spelling-only fix: it belongs in the evidence and correction log so another reviewer can see why a version changed.

For this case, conduct a three-part test. First, read the source without the target and write what is established, what is estimated and what remains unknown. Second, read only the target and ask whether a technically trained reader would reach the same conclusion. Third, compare the data-bearing elements—especially identity, time, quantity, status and authorization. If a real document does not contain enough information to close the case, state the missing field and return the question to the source owner.

42. Reentry mixed with collision

Symptom: Earth-impact casualty language used for orbital conjunction. Why it matters: Confuses hazard types. Repair: Separate risk domains. The safe review procedure is to compare the exact source assertion with the target assertion, identify which invariant was lost, and confirm that the replacement wording restores it. Do not treat this as a spelling-only fix: it belongs in the evidence and correction log so another reviewer can see why a version changed.

For this case, conduct a three-part test. First, read the source without the target and write what is established, what is estimated and what remains unknown. Second, read only the target and ask whether a technically trained reader would reach the same conclusion. Third, compare the data-bearing elements—especially identity, time, quantity, status and authorization. If a real document does not contain enough information to close the case, state the missing field and return the question to the source owner.

43. Observation called prediction

Symptom: Sensor track is narrated as forecast. Why it matters: Confuses evidence stage. Repair: Label raw, derived and propagated data. The safe review procedure is to compare the exact source assertion with the target assertion, identify which invariant was lost, and confirm that the replacement wording restores it. Do not treat this as a spelling-only fix: it belongs in the evidence and correction log so another reviewer can see why a version changed.

For this case, conduct a three-part test. First, read the source without the target and write what is established, what is estimated and what remains unknown. Second, read only the target and ask whether a technically trained reader would reach the same conclusion. Third, compare the data-bearing elements—especially identity, time, quantity, status and authorization. If a real document does not contain enough information to close the case, state the missing field and return the question to the source owner.

44. Prediction called observation

Symptom: Propagated state described as measured point. Why it matters: Overstates empirical status. Repair: Retain model dependency. The safe review procedure is to compare the exact source assertion with the target assertion, identify which invariant was lost, and confirm that the replacement wording restores it. Do not treat this as a spelling-only fix: it belongs in the evidence and correction log so another reviewer can see why a version changed.

For this case, conduct a three-part test. First, read the source without the target and write what is established, what is estimated and what remains unknown. Second, read only the target and ask whether a technically trained reader would reach the same conclusion. Third, compare the data-bearing elements—especially identity, time, quantity, status and authorization. If a real document does not contain enough information to close the case, state the missing field and return the question to the source owner.

45. Closure state omitted

Symptom: Old open warning republished after closure. Why it matters: Keeps a stale hazard alive. Repair: Publish status and correction date. The safe review procedure is to compare the exact source assertion with the target assertion, identify which invariant was lost, and confirm that the replacement wording restores it. Do not treat this as a spelling-only fix: it belongs in the evidence and correction log so another reviewer can see why a version changed.

For this case, conduct a three-part test. First, read the source without the target and write what is established, what is estimated and what remains unknown. Second, read only the target and ask whether a technically trained reader would reach the same conclusion. Third, compare the data-bearing elements—especially identity, time, quantity, status and authorization. If a real document does not contain enough information to close the case, state the missing field and return the question to the source owner.

46. Numerical range inverted

Symptom: Threshold written as above instead of below. Why it matters: Reverses interpretation. Repair: Validate comparison symbols and scope. The safe review procedure is to compare the exact source assertion with the target assertion, identify which invariant was lost, and confirm that the replacement wording restores it. Do not treat this as a spelling-only fix: it belongs in the evidence and correction log so another reviewer can see why a version changed.

For this case, conduct a three-part test. First, read the source without the target and write what is established, what is estimated and what remains unknown. Second, read only the target and ask whether a technically trained reader would reach the same conclusion. Third, compare the data-bearing elements—especially identity, time, quantity, status and authorization. If a real document does not contain enough information to close the case, state the missing field and return the question to the source owner.

47. Public explanation overclaims

Symptom: A headline says ‘satellite safe’. Why it matters: Turns conditional result into guarantee. Repair: Keep model caveats in snippet and headline. The safe review procedure is to compare the exact source assertion with the target assertion, identify which invariant was lost, and confirm that the replacement wording restores it. Do not treat this as a spelling-only fix: it belongs in the evidence and correction log so another reviewer can see why a version changed.

For this case, conduct a three-part test. First, read the source without the target and write what is established, what is estimated and what remains unknown. Second, read only the target and ask whether a technically trained reader would reach the same conclusion. Third, compare the data-bearing elements—especially identity, time, quantity, status and authorization. If a real document does not contain enough information to close the case, state the missing field and return the question to the source owner.

48. No independent technical reviewer

Symptom: Only prose quality assessed. Why it matters: Domain error reaches readers. Repair: Require qualified technical review. The safe review procedure is to compare the exact source assertion with the target assertion, identify which invariant was lost, and confirm that the replacement wording restores it. Do not treat this as a spelling-only fix: it belongs in the evidence and correction log so another reviewer can see why a version changed.

For this case, conduct a three-part test. First, read the source without the target and write what is established, what is estimated and what remains unknown. Second, read only the target and ask whether a technically trained reader would reach the same conclusion. Third, compare the data-bearing elements—especially identity, time, quantity, status and authorization. If a real document does not contain enough information to close the case, state the missing field and return the question to the source owner.

49. No human language review

Symptom: Only engineering correctness assessed. Why it matters: Target text remains ambiguous. Repair: Require qualified target-language reader. The safe review procedure is to compare the exact source assertion with the target assertion, identify which invariant was lost, and confirm that the replacement wording restores it. Do not treat this as a spelling-only fix: it belongs in the evidence and correction log so another reviewer can see why a version changed.

For this case, conduct a three-part test. First, read the source without the target and write what is established, what is estimated and what remains unknown. Second, read only the target and ask whether a technically trained reader would reach the same conclusion. Third, compare the data-bearing elements—especially identity, time, quantity, status and authorization. If a real document does not contain enough information to close the case, state the missing field and return the question to the source owner.

50. No retained source

Symptom: Snapshot and timestamp discarded. Why it matters: Cannot reconstruct correction. Repair: Archive input and release receipts. The safe review procedure is to compare the exact source assertion with the target assertion, identify which invariant was lost, and confirm that the replacement wording restores it. Do not treat this as a spelling-only fix: it belongs in the evidence and correction log so another reviewer can see why a version changed.

For this case, conduct a three-part test. First, read the source without the target and write what is established, what is estimated and what remains unknown. Second, read only the target and ask whether a technically trained reader would reach the same conclusion. Third, compare the data-bearing elements—especially identity, time, quantity, status and authorization. If a real document does not contain enough information to close the case, state the missing field and return the question to the source owner.

Sixty-two worked diagnostic cases: from broken wording to defensible meaning

The following scenarios are fictional teaching cases. The numbers and object labels are invented; the objective is to practice translating field relationships and uncertainty, not to recreate real warning products. For each case, read the source fact, locate the misleading interpretation, then test the corrected explanation against the evidence contract.

Worked case 01: Two objects, one event

Source fact: A screening note pairs protected object ORION-7 with another tracked object and labels the other ‘secondary’. Misleading target: A public summary calls secondary object a fragment of debris without source classification. The issue here is not whether the target sounds natural, but which independent fact it causes the reader to believe. A translator should first identify the value, the source of authority, the time reference and the degree of certainty before selecting a natural sentence. If any of those properties are missing from the provided record, the draft should disclose the absence rather than filling it with a confident guess.

Defensible rendering: Say ‘another tracked object’ until the catalogue identifies type; retain both source identifiers. Why the repair works: Identity is a record attribute, while primary/secondary is a role inside one analysis. Treat this as a role-and-reference problem. Identify every noun phrase that points to a real object, a model state, a report or a decision. The target must preserve those referents even if word order changes across languages. This is a practical exercise in reconstructing the same evidence in a different linguistic form. The source and target do not need identical syntax, but they must support the same conclusions and must not suggest a new mission action.

The diagnostic receipt for case 1 is small: retain the source phrase or field, record the rejected interpretation, state the revised wording, and indicate which source owner can resolve uncertainty. Then ask a second reader to explain the target without seeing the source. If they cannot distinguish prediction from observation, assessment from approval, or data availability from physical status where relevant, the revision is not finished. A public FAQ or search snippet derived from this case needs the same factual discipline as the full article.

Worked case 02: Close approach, not contact

Source fact: An analyst writes: a conjunction is predicted at 18:40 UTC, pending a fresh orbit update. Misleading target: The translation says two satellites will collide at 18:40. The issue here is not whether the target sounds natural, but which independent fact it causes the reader to believe. A translator should first identify the value, the source of authority, the time reference and the degree of certainty before selecting a natural sentence. If any of those properties are missing from the provided record, the draft should disclose the absence rather than filling it with a confident guess.

Defensible rendering: Translate as a predicted close approach at that time, with analysis pending. Why the repair works: The forecast did not assert collision and the subsequent orbit update may alter the result. Treat this as an epistemic-strength problem. Write down whether the source asserts an observation, an estimate, a prediction, a judgment or a completed act. The target must not silently move the claim to a stronger category. This is a practical exercise in reconstructing the same evidence in a different linguistic form. The source and target do not need identical syntax, but they must support the same conclusions and must not suggest a new mission action.

The diagnostic receipt for case 2 is small: retain the source phrase or field, record the rejected interpretation, state the revised wording, and indicate which source owner can resolve uncertainty. Then ask a second reader to explain the target without seeing the source. If they cannot distinguish prediction from observation, assessment from approval, or data availability from physical status where relevant, the revision is not finished. A public FAQ or search snippet derived from this case needs the same factual discipline as the full article.

Worked case 03: Two timestamps

Source fact: The header records generation at 03:12 UTC and TCA at 09:28 UTC. Misleading target: A diagram caption labels 03:12 as closest approach. The issue here is not whether the target sounds natural, but which independent fact it causes the reader to believe. A translator should first identify the value, the source of authority, the time reference and the degree of certainty before selecting a natural sentence. If any of those properties are missing from the provided record, the draft should disclose the absence rather than filling it with a confident guess.

Defensible rendering: Display both timestamps with different labels and no invented local timezone. Why the repair works: A message may be generated hours before the modeled event. Treat this as a numerical-invariant problem. Place the original token beside the rendered value, preserve any sign, unit, epoch or exponent, and run a back-conversion where applicable. Nice formatting cannot compensate for changed mathematics. This is a practical exercise in reconstructing the same evidence in a different linguistic form. The source and target do not need identical syntax, but they must support the same conclusions and must not suggest a new mission action.

The diagnostic receipt for case 3 is small: retain the source phrase or field, record the rejected interpretation, state the revised wording, and indicate which source owner can resolve uncertainty. Then ask a second reader to explain the target without seeing the source. If they cannot distinguish prediction from observation, assessment from approval, or data availability from physical status where relevant, the revision is not finished. A public FAQ or search snippet derived from this case needs the same factual discipline as the full article.

Worked case 04: Rollover after midnight

Source fact: TCA is 00:15 UTC on 2 November while a reader-facing conversion falls on the previous local date. Misleading target: The date is copied unchanged beside converted local time. The issue here is not whether the target sounds natural, but which independent fact it causes the reader to believe. A translator should first identify the value, the source of authority, the time reference and the degree of certainty before selecting a natural sentence. If any of those properties are missing from the provided record, the draft should disclose the absence rather than filling it with a confident guess.

Defensible rendering: Retain UTC and separately print the labeled local presentation date. Why the repair works: A date change is an expected consequence of conversion, not evidence of a revised event. Treat this as a provenance problem. Identify who issued the record, which version supplied the fact and how much of the statement comes from that source. Never assemble one authoritative-sounding sentence out of incompatible snapshots. This is a practical exercise in reconstructing the same evidence in a different linguistic form. The source and target do not need identical syntax, but they must support the same conclusions and must not suggest a new mission action.

The diagnostic receipt for case 4 is small: retain the source phrase or field, record the rejected interpretation, state the revised wording, and indicate which source owner can resolve uncertainty. Then ask a second reader to explain the target without seeing the source. If they cannot distinguish prediction from observation, assessment from approval, or data availability from physical status where relevant, the revision is not finished. A public FAQ or search snippet derived from this case needs the same factual discipline as the full article.

Worked case 05: Seconds and order

Source fact: Two revisions of a fictional alert have TCAs 12:00:02.7 and 12:00:03.4. Misleading target: Both are rounded to 12:00 before comparison. The issue here is not whether the target sounds natural, but which independent fact it causes the reader to believe. A translator should first identify the value, the source of authority, the time reference and the degree of certainty before selecting a natural sentence. If any of those properties are missing from the provided record, the draft should disclose the absence rather than filling it with a confident guess.

Defensible rendering: Keep raw timestamps in technical comparison and describe differences carefully. Why the repair works: Coarse rendering can hide a change even when a rounded public time is sufficient. Treat this as a time-line problem. Distinguish document creation, observation, prediction, review, decision and publication. A date can be syntactically correct while answering the wrong temporal question. This is a practical exercise in reconstructing the same evidence in a different linguistic form. The source and target do not need identical syntax, but they must support the same conclusions and must not suggest a new mission action.

The diagnostic receipt for case 5 is small: retain the source phrase or field, record the rejected interpretation, state the revised wording, and indicate which source owner can resolve uncertainty. Then ask a second reader to explain the target without seeing the source. If they cannot distinguish prediction from observation, assessment from approval, or data availability from physical status where relevant, the revision is not finished. A public FAQ or search snippet derived from this case needs the same factual discipline as the full article.

Worked case 06: Six-digit gap

Source fact: An operator searches for catalog 100021 using a legacy TLE-only display and gets no row. Misleading target: The guide says the object is no longer in orbit. The issue here is not whether the target sounds natural, but which independent fact it causes the reader to believe. A translator should first identify the value, the source of authority, the time reference and the degree of certainty before selecting a natural sentence. If any of those properties are missing from the provided record, the draft should disclose the absence rather than filling it with a confident guess.

Defensible rendering: Explain that the chosen delivery format lacks the object’s representable ID range. Why the repair works: Provider and schema limitations do not prove a physical event. Treat this as a machine-human boundary problem. A parser keyword or ID must stay exact; a narrative definition can be translated and clarified. A reviewer should inspect both representations rather than assuming that correct prose guarantees a valid data feed. This is a practical exercise in reconstructing the same evidence in a different linguistic form. The source and target do not need identical syntax, but they must support the same conclusions and must not suggest a new mission action.

The diagnostic receipt for case 6 is small: retain the source phrase or field, record the rejected interpretation, state the revised wording, and indicate which source owner can resolve uncertainty. Then ask a second reader to explain the target without seeing the source. If they cannot distinguish prediction from observation, assessment from approval, or data availability from physical status where relevant, the revision is not finished. A public FAQ or search snippet derived from this case needs the same factual discipline as the full article.

Worked case 07: Explicit GP format

Source fact: A GP query omits FORMAT after a feed’s documented default switches to CSV. Misleading target: A tutorial instructs readers to parse response as two-line elements. The issue here is not whether the target sounds natural, but which independent fact it causes the reader to believe. A translator should first identify the value, the source of authority, the time reference and the degree of certainty before selecting a natural sentence. If any of those properties are missing from the provided record, the draft should disclose the absence rather than filling it with a confident guess.

Defensible rendering: Require supported FORMAT selection and inspect the actual data schema. Why the repair works: A successful request alone does not demonstrate that a parser received the expected representation. Treat this as a reader-action problem. Imagine a non-specialist receiving only the translated summary. Would they infer a crash, an approved maneuver or a guarantee that the source never supplied? If so, the language needs a narrower assertion. This is a practical exercise in reconstructing the same evidence in a different linguistic form. The source and target do not need identical syntax, but they must support the same conclusions and must not suggest a new mission action.

The diagnostic receipt for case 7 is small: retain the source phrase or field, record the rejected interpretation, state the revised wording, and indicate which source owner can resolve uncertainty. Then ask a second reader to explain the target without seeing the source. If they cannot distinguish prediction from observation, assessment from approval, or data availability from physical status where relevant, the revision is not finished. A public FAQ or search snippet derived from this case needs the same factual discipline as the full article.

Worked case 08: Object alias

Source fact: A satellite is known by a mission nickname, spacecraft bus name and catalog number. Misleading target: The translation merges the nickname with a different deployed payload. The issue here is not whether the target sounds natural, but which independent fact it causes the reader to believe. A translator should first identify the value, the source of authority, the time reference and the degree of certainty before selecting a natural sentence. If any of those properties are missing from the provided record, the draft should disclose the absence rather than filling it with a confident guess.

Defensible rendering: Retain canonical identifier and list aliases with sources. Why the repair works: Display names are convenient for people but unsuitable as unique joins. Treat this as a correction-readiness problem. Record enough source data to repair the public explanation after a future message changes the event. A source snapshot and a dated release note are more useful than an unlabeled copied paragraph. This is a practical exercise in reconstructing the same evidence in a different linguistic form. The source and target do not need identical syntax, but they must support the same conclusions and must not suggest a new mission action.

The diagnostic receipt for case 8 is small: retain the source phrase or field, record the rejected interpretation, state the revised wording, and indicate which source owner can resolve uncertainty. Then ask a second reader to explain the target without seeing the source. If they cannot distinguish prediction from observation, assessment from approval, or data availability from physical status where relevant, the revision is not finished. A public FAQ or search snippet derived from this case needs the same factual discipline as the full article.

Worked case 09: Scientific notation

Source fact: An estimated event probability is printed as 3.0E-06. Misleading target: The translation says three chances in a hundred thousand. The issue here is not whether the target sounds natural, but which independent fact it causes the reader to believe. A translator should first identify the value, the source of authority, the time reference and the degree of certainty before selecting a natural sentence. If any of those properties are missing from the provided record, the draft should disclose the absence rather than filling it with a confident guess.

Defensible rendering: Preserve 3.0E-06; if explaining, note that it is three per million under that model. Why the repair works: Exponent interpretation is a frequent risk in numbers intended for public readers. Treat this as a role-and-reference problem. Identify every noun phrase that points to a real object, a model state, a report or a decision. The target must preserve those referents even if word order changes across languages. This is a practical exercise in reconstructing the same evidence in a different linguistic form. The source and target do not need identical syntax, but they must support the same conclusions and must not suggest a new mission action.

The diagnostic receipt for case 9 is small: retain the source phrase or field, record the rejected interpretation, state the revised wording, and indicate which source owner can resolve uncertainty. Then ask a second reader to explain the target without seeing the source. If they cannot distinguish prediction from observation, assessment from approval, or data availability from physical status where relevant, the revision is not finished. A public FAQ or search snippet derived from this case needs the same factual discipline as the full article.

Worked case 10: Probability not certainty

Source fact: A model reports a small estimated collision probability for one encounter. Misleading target: The summary promises that collision is impossible. The issue here is not whether the target sounds natural, but which independent fact it causes the reader to believe. A translator should first identify the value, the source of authority, the time reference and the degree of certainty before selecting a natural sentence. If any of those properties are missing from the provided record, the draft should disclose the absence rather than filling it with a confident guess.

Defensible rendering: Say the model’s estimated probability is low without asserting a guarantee. Why the repair works: A conditional estimate is not an omniscient judgment. Treat this as an epistemic-strength problem. Write down whether the source asserts an observation, an estimate, a prediction, a judgment or a completed act. The target must not silently move the claim to a stronger category. This is a practical exercise in reconstructing the same evidence in a different linguistic form. The source and target do not need identical syntax, but they must support the same conclusions and must not suggest a new mission action.

The diagnostic receipt for case 10 is small: retain the source phrase or field, record the rejected interpretation, state the revised wording, and indicate which source owner can resolve uncertainty. Then ask a second reader to explain the target without seeing the source. If they cannot distinguish prediction from observation, assessment from approval, or data availability from physical status where relevant, the revision is not finished. A public FAQ or search snippet derived from this case needs the same factual discipline as the full article.

Worked case 11: Missing Pc

Source fact: The collision-probability field is absent while a miss distance is supplied. Misleading target: The reviewer inserts zero for an empty cell. The issue here is not whether the target sounds natural, but which independent fact it causes the reader to believe. A translator should first identify the value, the source of authority, the time reference and the degree of certainty before selecting a natural sentence. If any of those properties are missing from the provided record, the draft should disclose the absence rather than filling it with a confident guess.

Defensible rendering: Display ‘not reported’ and seek authoritative clarification if probability is required. Why the repair works: Zero and missing are distinct states in both mathematics and data governance. Treat this as a numerical-invariant problem. Place the original token beside the rendered value, preserve any sign, unit, epoch or exponent, and run a back-conversion where applicable. Nice formatting cannot compensate for changed mathematics. This is a practical exercise in reconstructing the same evidence in a different linguistic form. The source and target do not need identical syntax, but they must support the same conclusions and must not suggest a new mission action.

The diagnostic receipt for case 11 is small: retain the source phrase or field, record the rejected interpretation, state the revised wording, and indicate which source owner can resolve uncertainty. Then ask a second reader to explain the target without seeing the source. If they cannot distinguish prediction from observation, assessment from approval, or data availability from physical status where relevant, the revision is not finished. A public FAQ or search snippet derived from this case needs the same factual discipline as the full article.

Worked case 12: Miss distance

Source fact: A report estimates a minimum separation of 860 m at TCA. Misleading target: Public copy says the satellites are currently 860 m apart. The issue here is not whether the target sounds natural, but which independent fact it causes the reader to believe. A translator should first identify the value, the source of authority, the time reference and the degree of certainty before selecting a natural sentence. If any of those properties are missing from the provided record, the draft should disclose the absence rather than filling it with a confident guess.

Defensible rendering: Use predicted minimum separation of 860 m at the specified TCA. Why the repair works: Measurement time and prediction status determine what a distance refers to. Treat this as a provenance problem. Identify who issued the record, which version supplied the fact and how much of the statement comes from that source. Never assemble one authoritative-sounding sentence out of incompatible snapshots. This is a practical exercise in reconstructing the same evidence in a different linguistic form. The source and target do not need identical syntax, but they must support the same conclusions and must not suggest a new mission action.

The diagnostic receipt for case 12 is small: retain the source phrase or field, record the rejected interpretation, state the revised wording, and indicate which source owner can resolve uncertainty. Then ask a second reader to explain the target without seeing the source. If they cannot distinguish prediction from observation, assessment from approval, or data availability from physical status where relevant, the revision is not finished. A public FAQ or search snippet derived from this case needs the same factual discipline as the full article.

Worked case 13: Relative speed

Source fact: The source calls 12.1 km/s the predicted relative speed. Misleading target: Translated text calls it ORION-7’s orbital speed. The issue here is not whether the target sounds natural, but which independent fact it causes the reader to believe. A translator should first identify the value, the source of authority, the time reference and the degree of certainty before selecting a natural sentence. If any of those properties are missing from the provided record, the draft should disclose the absence rather than filling it with a confident guess.

Defensible rendering: Name the speed between the objects at the encounter. Why the repair works: Relative and absolute velocities answer different physical questions. Treat this as a time-line problem. Distinguish document creation, observation, prediction, review, decision and publication. A date can be syntactically correct while answering the wrong temporal question. This is a practical exercise in reconstructing the same evidence in a different linguistic form. The source and target do not need identical syntax, but they must support the same conclusions and must not suggest a new mission action.

The diagnostic receipt for case 13 is small: retain the source phrase or field, record the rejected interpretation, state the revised wording, and indicate which source owner can resolve uncertainty. Then ask a second reader to explain the target without seeing the source. If they cannot distinguish prediction from observation, assessment from approval, or data availability from physical status where relevant, the revision is not finished. A public FAQ or search snippet derived from this case needs the same factual discipline as the full article.

Worked case 14: Signed radial component

Source fact: A source field stores a radial relative-position component of -45 m. Misleading target: The minus sign is omitted in a formatted table. The issue here is not whether the target sounds natural, but which independent fact it causes the reader to believe. A translator should first identify the value, the source of authority, the time reference and the degree of certainty before selecting a natural sentence. If any of those properties are missing from the provided record, the draft should disclose the absence rather than filling it with a confident guess.

Defensible rendering: Preserve -45 m and identify the RTN convention. Why the repair works: Signs are part of the geometry, not editorial punctuation. Treat this as a machine-human boundary problem. A parser keyword or ID must stay exact; a narrative definition can be translated and clarified. A reviewer should inspect both representations rather than assuming that correct prose guarantees a valid data feed. This is a practical exercise in reconstructing the same evidence in a different linguistic form. The source and target do not need identical syntax, but they must support the same conclusions and must not suggest a new mission action.

The diagnostic receipt for case 14 is small: retain the source phrase or field, record the rejected interpretation, state the revised wording, and indicate which source owner can resolve uncertainty. Then ask a second reader to explain the target without seeing the source. If they cannot distinguish prediction from observation, assessment from approval, or data availability from physical status where relevant, the revision is not finished. A public FAQ or search snippet derived from this case needs the same factual discipline as the full article.

Worked case 15: Frame label lost

Source fact: A matrix is specified in RTN coordinates. Misleading target: A caption says coordinates are east, north and up. The issue here is not whether the target sounds natural, but which independent fact it causes the reader to believe. A translator should first identify the value, the source of authority, the time reference and the degree of certainty before selecting a natural sentence. If any of those properties are missing from the provided record, the draft should disclose the absence rather than filling it with a confident guess.

Defensible rendering: Keep RTN or explain radial, transverse and normal relative to orbit geometry. Why the repair works: Ordinary compass labels imply a different frame. Treat this as a reader-action problem. Imagine a non-specialist receiving only the translated summary. Would they infer a crash, an approved maneuver or a guarantee that the source never supplied? If so, the language needs a narrower assertion. This is a practical exercise in reconstructing the same evidence in a different linguistic form. The source and target do not need identical syntax, but they must support the same conclusions and must not suggest a new mission action.

The diagnostic receipt for case 15 is small: retain the source phrase or field, record the rejected interpretation, state the revised wording, and indicate which source owner can resolve uncertainty. Then ask a second reader to explain the target without seeing the source. If they cannot distinguish prediction from observation, assessment from approval, or data availability from physical status where relevant, the revision is not finished. A public FAQ or search snippet derived from this case needs the same factual discipline as the full article.

Worked case 16: Matrix meaning

Source fact: A CDM includes several position covariance entries. Misleading target: A digest calls the largest entry ‘location uncertainty in metres’. The issue here is not whether the target sounds natural, but which independent fact it causes the reader to believe. A translator should first identify the value, the source of authority, the time reference and the degree of certainty before selecting a natural sentence. If any of those properties are missing from the provided record, the draft should disclose the absence rather than filling it with a confident guess.

Defensible rendering: Describe the matrix as covariance and let qualified analysis determine a derived uncertainty summary. Why the repair works: Covariance entries have dimensional and correlation meanings. Treat this as a correction-readiness problem. Record enough source data to repair the public explanation after a future message changes the event. A source snapshot and a dated release note are more useful than an unlabeled copied paragraph. This is a practical exercise in reconstructing the same evidence in a different linguistic form. The source and target do not need identical syntax, but they must support the same conclusions and must not suggest a new mission action.

The diagnostic receipt for case 16 is small: retain the source phrase or field, record the rejected interpretation, state the revised wording, and indicate which source owner can resolve uncertainty. Then ask a second reader to explain the target without seeing the source. If they cannot distinguish prediction from observation, assessment from approval, or data availability from physical status where relevant, the revision is not finished. A public FAQ or search snippet derived from this case needs the same factual discipline as the full article.

Worked case 17: Correlation dropped

Source fact: Off-diagonal covariance fields are present in the source. Misleading target: Only three diagonal entries are kept because a table seems crowded. The issue here is not whether the target sounds natural, but which independent fact it causes the reader to believe. A translator should first identify the value, the source of authority, the time reference and the degree of certainty before selecting a natural sentence. If any of those properties are missing from the provided record, the draft should disclose the absence rather than filling it with a confident guess.

Defensible rendering: Retain complete machine data and label any separate visual simplification. Why the repair works: Removing cross terms can alter mathematical interpretation. Treat this as a role-and-reference problem. Identify every noun phrase that points to a real object, a model state, a report or a decision. The target must preserve those referents even if word order changes across languages. This is a practical exercise in reconstructing the same evidence in a different linguistic form. The source and target do not need identical syntax, but they must support the same conclusions and must not suggest a new mission action.

The diagnostic receipt for case 17 is small: retain the source phrase or field, record the rejected interpretation, state the revised wording, and indicate which source owner can resolve uncertainty. Then ask a second reader to explain the target without seeing the source. If they cannot distinguish prediction from observation, assessment from approval, or data availability from physical status where relevant, the revision is not finished. A public FAQ or search snippet derived from this case needs the same factual discipline as the full article.

Worked case 18: Different covariance epochs

Source fact: Two messages have uncertainty representations tied to distinct orbit-solution epochs. Misleading target: A comparison chart assumes their values share the same time base. The issue here is not whether the target sounds natural, but which independent fact it causes the reader to believe. A translator should first identify the value, the source of authority, the time reference and the degree of certainty before selecting a natural sentence. If any of those properties are missing from the provided record, the draft should disclose the absence rather than filling it with a confident guess.

Defensible rendering: Label epochs and only compare values under an explicitly justified procedure. Why the repair works: Model age affects how a value may be interpreted. Treat this as an epistemic-strength problem. Write down whether the source asserts an observation, an estimate, a prediction, a judgment or a completed act. The target must not silently move the claim to a stronger category. This is a practical exercise in reconstructing the same evidence in a different linguistic form. The source and target do not need identical syntax, but they must support the same conclusions and must not suggest a new mission action.

The diagnostic receipt for case 18 is small: retain the source phrase or field, record the rejected interpretation, state the revised wording, and indicate which source owner can resolve uncertainty. Then ask a second reader to explain the target without seeing the source. If they cannot distinguish prediction from observation, assessment from approval, or data availability from physical status where relevant, the revision is not finished. A public FAQ or search snippet derived from this case needs the same factual discipline as the full article.

Worked case 19: Hard-body radius

Source fact: An assumed combined hard-body radius is listed in analysis metadata. Misleading target: Public text says the satellite’s physical diameter was directly measured. The issue here is not whether the target sounds natural, but which independent fact it causes the reader to believe. A translator should first identify the value, the source of authority, the time reference and the degree of certainty before selecting a natural sentence. If any of those properties are missing from the provided record, the draft should disclose the absence rather than filling it with a confident guess.

Defensible rendering: Describe the stated collision-geometry assumption, not an observed size. Why the repair works: Inputs to a model are not necessarily sensor measurements. Treat this as a numerical-invariant problem. Place the original token beside the rendered value, preserve any sign, unit, epoch or exponent, and run a back-conversion where applicable. Nice formatting cannot compensate for changed mathematics. This is a practical exercise in reconstructing the same evidence in a different linguistic form. The source and target do not need identical syntax, but they must support the same conclusions and must not suggest a new mission action.

The diagnostic receipt for case 19 is small: retain the source phrase or field, record the rejected interpretation, state the revised wording, and indicate which source owner can resolve uncertainty. Then ask a second reader to explain the target without seeing the source. If they cannot distinguish prediction from observation, assessment from approval, or data availability from physical status where relevant, the revision is not finished. A public FAQ or search snippet derived from this case needs the same factual discipline as the full article.

Worked case 20: Change in Pc

Source fact: A later message shows a higher probability after more tracking data arrive. Misleading target: A headline says the two objects moved physically closer. The issue here is not whether the target sounds natural, but which independent fact it causes the reader to believe. A translator should first identify the value, the source of authority, the time reference and the degree of certainty before selecting a natural sentence. If any of those properties are missing from the provided record, the draft should disclose the absence rather than filling it with a confident guess.

Defensible rendering: Describe what the analysis changed; do not infer a geometric cause without evidence. Why the repair works: Probability changes can reflect uncertainty estimates, trajectories or model settings. Treat this as a provenance problem. Identify who issued the record, which version supplied the fact and how much of the statement comes from that source. Never assemble one authoritative-sounding sentence out of incompatible snapshots. This is a practical exercise in reconstructing the same evidence in a different linguistic form. The source and target do not need identical syntax, but they must support the same conclusions and must not suggest a new mission action.

The diagnostic receipt for case 20 is small: retain the source phrase or field, record the rejected interpretation, state the revised wording, and indicate which source owner can resolve uncertainty. Then ask a second reader to explain the target without seeing the source. If they cannot distinguish prediction from observation, assessment from approval, or data availability from physical status where relevant, the revision is not finished. A public FAQ or search snippet derived from this case needs the same factual discipline as the full article.

Worked case 21: Change in miss distance

Source fact: A later solution gives a different predicted miss distance for the same event. Misleading target: A glossary treats both values as simultaneous direct measurements. The issue here is not whether the target sounds natural, but which independent fact it causes the reader to believe. A translator should first identify the value, the source of authority, the time reference and the degree of certainty before selecting a natural sentence. If any of those properties are missing from the provided record, the draft should disclose the absence rather than filling it with a confident guess.

Defensible rendering: Present each as a versioned prediction and keep the generation time. Why the repair works: Forecast revisions are a feature of estimation, not contradictory physical observations. Treat this as a time-line problem. Distinguish document creation, observation, prediction, review, decision and publication. A date can be syntactically correct while answering the wrong temporal question. This is a practical exercise in reconstructing the same evidence in a different linguistic form. The source and target do not need identical syntax, but they must support the same conclusions and must not suggest a new mission action.

The diagnostic receipt for case 21 is small: retain the source phrase or field, record the rejected interpretation, state the revised wording, and indicate which source owner can resolve uncertainty. Then ask a second reader to explain the target without seeing the source. If they cannot distinguish prediction from observation, assessment from approval, or data availability from physical status where relevant, the revision is not finished. A public FAQ or search snippet derived from this case needs the same factual discipline as the full article.

Worked case 22: Screening threshold

Source fact: A system records that the pair entered a configured screening volume. Misleading target: The public release calls it an emergency-level risk. The issue here is not whether the target sounds natural, but which independent fact it causes the reader to believe. A translator should first identify the value, the source of authority, the time reference and the degree of certainty before selecting a natural sentence. If any of those properties are missing from the provided record, the draft should disclose the absence rather than filling it with a confident guess.

Defensible rendering: Say it met screening criteria and was sent for assessment. Why the repair works: Selection for review is distinct from assessment policy. Treat this as a machine-human boundary problem. A parser keyword or ID must stay exact; a narrative definition can be translated and clarified. A reviewer should inspect both representations rather than assuming that correct prose guarantees a valid data feed. This is a practical exercise in reconstructing the same evidence in a different linguistic form. The source and target do not need identical syntax, but they must support the same conclusions and must not suggest a new mission action.

The diagnostic receipt for case 22 is small: retain the source phrase or field, record the rejected interpretation, state the revised wording, and indicate which source owner can resolve uncertainty. Then ask a second reader to explain the target without seeing the source. If they cannot distinguish prediction from observation, assessment from approval, or data availability from physical status where relevant, the revision is not finished. A public FAQ or search snippet derived from this case needs the same factual discipline as the full article.

Worked case 23: Alert color

Source fact: A provider uses amber as a local internal category. Misleading target: The translation assigns a universal international amber-alert meaning. The issue here is not whether the target sounds natural, but which independent fact it causes the reader to believe. A translator should first identify the value, the source of authority, the time reference and the degree of certainty before selecting a natural sentence. If any of those properties are missing from the provided record, the draft should disclose the absence rather than filling it with a confident guess.

Defensible rendering: Retain the provider’s exact scheme and avoid generic equivalence. Why the repair works: Color vocabularies do not make policy categories interoperable on their own. Treat this as a reader-action problem. Imagine a non-specialist receiving only the translated summary. Would they infer a crash, an approved maneuver or a guarantee that the source never supplied? If so, the language needs a narrower assertion. This is a practical exercise in reconstructing the same evidence in a different linguistic form. The source and target do not need identical syntax, but they must support the same conclusions and must not suggest a new mission action.

The diagnostic receipt for case 23 is small: retain the source phrase or field, record the rejected interpretation, state the revised wording, and indicate which source owner can resolve uncertainty. Then ask a second reader to explain the target without seeing the source. If they cannot distinguish prediction from observation, assessment from approval, or data availability from physical status where relevant, the revision is not finished. A public FAQ or search snippet derived from this case needs the same factual discipline as the full article.

Worked case 24: Observed versus estimated

Source fact: A track is derived from sensor observations using an orbit-determination model. Misleading target: The article says every state vector component was directly measured. The issue here is not whether the target sounds natural, but which independent fact it causes the reader to believe. A translator should first identify the value, the source of authority, the time reference and the degree of certainty before selecting a natural sentence. If any of those properties are missing from the provided record, the draft should disclose the absence rather than filling it with a confident guess.

Defensible rendering: Write ‘estimated orbit state derived from observations’. Why the repair works: Model-derived values deserve their own provenance. Treat this as a correction-readiness problem. Record enough source data to repair the public explanation after a future message changes the event. A source snapshot and a dated release note are more useful than an unlabeled copied paragraph. This is a practical exercise in reconstructing the same evidence in a different linguistic form. The source and target do not need identical syntax, but they must support the same conclusions and must not suggest a new mission action.

The diagnostic receipt for case 24 is small: retain the source phrase or field, record the rejected interpretation, state the revised wording, and indicate which source owner can resolve uncertainty. Then ask a second reader to explain the target without seeing the source. If they cannot distinguish prediction from observation, assessment from approval, or data availability from physical status where relevant, the revision is not finished. A public FAQ or search snippet derived from this case needs the same factual discipline as the full article.

Worked case 25: TLE epoch

Source fact: A legacy element set contains an epoch older than its download date. Misleading target: The translated guide says the orbit was measured this morning. The issue here is not whether the target sounds natural, but which independent fact it causes the reader to believe. A translator should first identify the value, the source of authority, the time reference and the degree of certainty before selecting a natural sentence. If any of those properties are missing from the provided record, the draft should disclose the absence rather than filling it with a confident guess.

Defensible rendering: Show both epoch and retrieval time. Why the repair works: File freshness and underlying observation currency are not the same. Treat this as a role-and-reference problem. Identify every noun phrase that points to a real object, a model state, a report or a decision. The target must preserve those referents even if word order changes across languages. This is a practical exercise in reconstructing the same evidence in a different linguistic form. The source and target do not need identical syntax, but they must support the same conclusions and must not suggest a new mission action.

The diagnostic receipt for case 25 is small: retain the source phrase or field, record the rejected interpretation, state the revised wording, and indicate which source owner can resolve uncertainty. Then ask a second reader to explain the target without seeing the source. If they cannot distinguish prediction from observation, assessment from approval, or data availability from physical status where relevant, the revision is not finished. A public FAQ or search snippet derived from this case needs the same factual discipline as the full article.

Worked case 26: OMM field tokens

Source fact: OMM KVN contains an OBJECT_ID field and a plain-language editorial comment. Misleading target: Both field name and comment are translated into the target language. The issue here is not whether the target sounds natural, but which independent fact it causes the reader to believe. A translator should first identify the value, the source of authority, the time reference and the degree of certainty before selecting a natural sentence. If any of those properties are missing from the provided record, the draft should disclose the absence rather than filling it with a confident guess.

Defensible rendering: Keep OBJECT_ID unchanged and translate the comment only. Why the repair works: Schemas need stable key names even when explanatory language changes. Treat this as an epistemic-strength problem. Write down whether the source asserts an observation, an estimate, a prediction, a judgment or a completed act. The target must not silently move the claim to a stronger category. This is a practical exercise in reconstructing the same evidence in a different linguistic form. The source and target do not need identical syntax, but they must support the same conclusions and must not suggest a new mission action.

The diagnostic receipt for case 26 is small: retain the source phrase or field, record the rejected interpretation, state the revised wording, and indicate which source owner can resolve uncertainty. Then ask a second reader to explain the target without seeing the source. If they cannot distinguish prediction from observation, assessment from approval, or data availability from physical status where relevant, the revision is not finished. A public FAQ or search snippet derived from this case needs the same factual discipline as the full article.

Worked case 27: CSV delimiter

Source fact: An orbital catalogue exports comma-separated data, while the target locale uses decimal commas. Misleading target: Numbers split incorrectly when imported into a spreadsheet. The issue here is not whether the target sounds natural, but which independent fact it causes the reader to believe. A translator should first identify the value, the source of authority, the time reference and the degree of certainty before selecting a natural sentence. If any of those properties are missing from the provided record, the draft should disclose the absence rather than filling it with a confident guess.

Defensible rendering: Use an explicit parsing and locale rule without altering raw data. Why the repair works: Serialization and numeric punctuation must be coordinated. Treat this as a numerical-invariant problem. Place the original token beside the rendered value, preserve any sign, unit, epoch or exponent, and run a back-conversion where applicable. Nice formatting cannot compensate for changed mathematics. This is a practical exercise in reconstructing the same evidence in a different linguistic form. The source and target do not need identical syntax, but they must support the same conclusions and must not suggest a new mission action.

The diagnostic receipt for case 27 is small: retain the source phrase or field, record the rejected interpretation, state the revised wording, and indicate which source owner can resolve uncertainty. Then ask a second reader to explain the target without seeing the source. If they cannot distinguish prediction from observation, assessment from approval, or data availability from physical status where relevant, the revision is not finished. A public FAQ or search snippet derived from this case needs the same factual discipline as the full article.

Worked case 28: RTL identifiers

Source fact: A right-to-left paragraph embeds an ASCII catalog number and E-notation probability. Misleading target: Punctuation and digits display in a confusing order. The issue here is not whether the target sounds natural, but which independent fact it causes the reader to believe. A translator should first identify the value, the source of authority, the time reference and the degree of certainty before selecting a natural sentence. If any of those properties are missing from the provided record, the draft should disclose the absence rather than filling it with a confident guess.

Defensible rendering: Isolate machine tokens and verify visual and copied values. Why the repair works: Readable display can diverge from the underlying byte sequence. Treat this as a provenance problem. Identify who issued the record, which version supplied the fact and how much of the statement comes from that source. Never assemble one authoritative-sounding sentence out of incompatible snapshots. This is a practical exercise in reconstructing the same evidence in a different linguistic form. The source and target do not need identical syntax, but they must support the same conclusions and must not suggest a new mission action.

The diagnostic receipt for case 28 is small: retain the source phrase or field, record the rejected interpretation, state the revised wording, and indicate which source owner can resolve uncertainty. Then ask a second reader to explain the target without seeing the source. If they cannot distinguish prediction from observation, assessment from approval, or data availability from physical status where relevant, the revision is not finished. A public FAQ or search snippet derived from this case needs the same factual discipline as the full article.

Worked case 29: Provider disagreement

Source fact: Two data providers report differing TCAs for the same object pair. Misleading target: The editor averages their times and presents one answer. The issue here is not whether the target sounds natural, but which independent fact it causes the reader to believe. A translator should first identify the value, the source of authority, the time reference and the degree of certainty before selecting a natural sentence. If any of those properties are missing from the provided record, the draft should disclose the absence rather than filling it with a confident guess.

Defensible rendering: Keep separate sourced predictions and refer reconciliation to qualified analysts. Why the repair works: Different models and vintages must not be merged by editorial intuition. Treat this as a time-line problem. Distinguish document creation, observation, prediction, review, decision and publication. A date can be syntactically correct while answering the wrong temporal question. This is a practical exercise in reconstructing the same evidence in a different linguistic form. The source and target do not need identical syntax, but they must support the same conclusions and must not suggest a new mission action.

The diagnostic receipt for case 29 is small: retain the source phrase or field, record the rejected interpretation, state the revised wording, and indicate which source owner can resolve uncertainty. Then ask a second reader to explain the target without seeing the source. If they cannot distinguish prediction from observation, assessment from approval, or data availability from physical status where relevant, the revision is not finished. A public FAQ or search snippet derived from this case needs the same factual discipline as the full article.

Worked case 30: Stale alert

Source fact: An early risk alert is followed by a source notice marked closed. Misleading target: Search-index snippet continues to state the original risk as current. The issue here is not whether the target sounds natural, but which independent fact it causes the reader to believe. A translator should first identify the value, the source of authority, the time reference and the degree of certainty before selecting a natural sentence. If any of those properties are missing from the provided record, the draft should disclose the absence rather than filling it with a confident guess.

Defensible rendering: Mark previous version superseded and update the snippet with correction context. Why the repair works: An otherwise accurate old statement becomes misleading if its status is hidden. Treat this as a machine-human boundary problem. A parser keyword or ID must stay exact; a narrative definition can be translated and clarified. A reviewer should inspect both representations rather than assuming that correct prose guarantees a valid data feed. This is a practical exercise in reconstructing the same evidence in a different linguistic form. The source and target do not need identical syntax, but they must support the same conclusions and must not suggest a new mission action.

The diagnostic receipt for case 30 is small: retain the source phrase or field, record the rejected interpretation, state the revised wording, and indicate which source owner can resolve uncertainty. Then ask a second reader to explain the target without seeing the source. If they cannot distinguish prediction from observation, assessment from approval, or data availability from physical status where relevant, the revision is not finished. A public FAQ or search snippet derived from this case needs the same factual discipline as the full article.

Worked case 31: Recommendation authority

Source fact: A screening service advises that the event deserves operator attention. Misleading target: The translation says a spacecraft must execute a maneuver. The issue here is not whether the target sounds natural, but which independent fact it causes the reader to believe. A translator should first identify the value, the source of authority, the time reference and the degree of certainty before selecting a natural sentence. If any of those properties are missing from the provided record, the draft should disclose the absence rather than filling it with a confident guess.

Defensible rendering: Retain advisory status and identify who can authorize actions. Why the repair works: An educational text must not invent flight commands. Treat this as a reader-action problem. Imagine a non-specialist receiving only the translated summary. Would they infer a crash, an approved maneuver or a guarantee that the source never supplied? If so, the language needs a narrower assertion. This is a practical exercise in reconstructing the same evidence in a different linguistic form. The source and target do not need identical syntax, but they must support the same conclusions and must not suggest a new mission action.

The diagnostic receipt for case 31 is small: retain the source phrase or field, record the rejected interpretation, state the revised wording, and indicate which source owner can resolve uncertainty. Then ask a second reader to explain the target without seeing the source. If they cannot distinguish prediction from observation, assessment from approval, or data availability from physical status where relevant, the revision is not finished. A public FAQ or search snippet derived from this case needs the same factual discipline as the full article.

Worked case 32: Planned versus executed

Source fact: A flight note says a maneuver is planned but not yet approved. Misleading target: A multilingual timeline marks it performed. The issue here is not whether the target sounds natural, but which independent fact it causes the reader to believe. A translator should first identify the value, the source of authority, the time reference and the degree of certainty before selecting a natural sentence. If any of those properties are missing from the provided record, the draft should disclose the absence rather than filling it with a confident guess.

Defensible rendering: Keep ‘planned’ and include later execution only if a new source confirms it. Why the repair works: Planning, commanding, executing and verifying are distinct facts. Treat this as a correction-readiness problem. Record enough source data to repair the public explanation after a future message changes the event. A source snapshot and a dated release note are more useful than an unlabeled copied paragraph. This is a practical exercise in reconstructing the same evidence in a different linguistic form. The source and target do not need identical syntax, but they must support the same conclusions and must not suggest a new mission action.

The diagnostic receipt for case 32 is small: retain the source phrase or field, record the rejected interpretation, state the revised wording, and indicate which source owner can resolve uncertainty. Then ask a second reader to explain the target without seeing the source. If they cannot distinguish prediction from observation, assessment from approval, or data availability from physical status where relevant, the revision is not finished. A public FAQ or search snippet derived from this case needs the same factual discipline as the full article.

Worked case 33: No burn approval

Source fact: A translated analyst note contains ‘maneuver not approved’. Misleading target: Negation is lost and readers see ‘maneuver approved’. The issue here is not whether the target sounds natural, but which independent fact it causes the reader to believe. A translator should first identify the value, the source of authority, the time reference and the degree of certainty before selecting a natural sentence. If any of those properties are missing from the provided record, the draft should disclose the absence rather than filling it with a confident guess.

Defensible rendering: Make the negative status prominent and independently verify it. Why the repair works: A missing ‘not’ is a decision-state inversion. Treat this as a role-and-reference problem. Identify every noun phrase that points to a real object, a model state, a report or a decision. The target must preserve those referents even if word order changes across languages. This is a practical exercise in reconstructing the same evidence in a different linguistic form. The source and target do not need identical syntax, but they must support the same conclusions and must not suggest a new mission action.

The diagnostic receipt for case 33 is small: retain the source phrase or field, record the rejected interpretation, state the revised wording, and indicate which source owner can resolve uncertainty. Then ask a second reader to explain the target without seeing the source. If they cannot distinguish prediction from observation, assessment from approval, or data availability from physical status where relevant, the revision is not finished. A public FAQ or search snippet derived from this case needs the same factual discipline as the full article.

Worked case 34: False safety promise

Source fact: A low-risk screening outcome is summarized as ‘there is no danger’. Misleading target: Readers receive an absolute guarantee. The issue here is not whether the target sounds natural, but which independent fact it causes the reader to believe. A translator should first identify the value, the source of authority, the time reference and the degree of certainty before selecting a natural sentence. If any of those properties are missing from the provided record, the draft should disclose the absence rather than filling it with a confident guess.

Defensible rendering: State that current assessment did not meet a particular risk criterion under specified inputs. Why the repair works: A policy decision is contextual, not proof of zero possible risk. Treat this as an epistemic-strength problem. Write down whether the source asserts an observation, an estimate, a prediction, a judgment or a completed act. The target must not silently move the claim to a stronger category. This is a practical exercise in reconstructing the same evidence in a different linguistic form. The source and target do not need identical syntax, but they must support the same conclusions and must not suggest a new mission action.

The diagnostic receipt for case 34 is small: retain the source phrase or field, record the rejected interpretation, state the revised wording, and indicate which source owner can resolve uncertainty. Then ask a second reader to explain the target without seeing the source. If they cannot distinguish prediction from observation, assessment from approval, or data availability from physical status where relevant, the revision is not finished. A public FAQ or search snippet derived from this case needs the same factual discipline as the full article.

Worked case 35: Relative geometry caption

Source fact: A schematic exaggerates separation for visibility and carries ‘not to scale’. Misleading target: Target caption omits the qualification. The issue here is not whether the target sounds natural, but which independent fact it causes the reader to believe. A translator should first identify the value, the source of authority, the time reference and the degree of certainty before selecting a natural sentence. If any of those properties are missing from the provided record, the draft should disclose the absence rather than filling it with a confident guess.

Defensible rendering: Keep ‘not to scale’ and avoid deriving distances from the drawing. Why the repair works: Visualization can aid comprehension without serving as measured evidence. Treat this as a numerical-invariant problem. Place the original token beside the rendered value, preserve any sign, unit, epoch or exponent, and run a back-conversion where applicable. Nice formatting cannot compensate for changed mathematics. This is a practical exercise in reconstructing the same evidence in a different linguistic form. The source and target do not need identical syntax, but they must support the same conclusions and must not suggest a new mission action.

The diagnostic receipt for case 35 is small: retain the source phrase or field, record the rejected interpretation, state the revised wording, and indicate which source owner can resolve uncertainty. Then ask a second reader to explain the target without seeing the source. If they cannot distinguish prediction from observation, assessment from approval, or data availability from physical status where relevant, the revision is not finished. A public FAQ or search snippet derived from this case needs the same factual discipline as the full article.

Worked case 36: Probability chart axis

Source fact: A logarithmic plot displays values across several orders of magnitude. Misleading target: A localized caption assumes a linear scale. The issue here is not whether the target sounds natural, but which independent fact it causes the reader to believe. A translator should first identify the value, the source of authority, the time reference and the degree of certainty before selecting a natural sentence. If any of those properties are missing from the provided record, the draft should disclose the absence rather than filling it with a confident guess.

Defensible rendering: Describe the logarithmic axis and preserve tick labels. Why the repair works: Scale interpretation changes the perceived trend. Treat this as a provenance problem. Identify who issued the record, which version supplied the fact and how much of the statement comes from that source. Never assemble one authoritative-sounding sentence out of incompatible snapshots. This is a practical exercise in reconstructing the same evidence in a different linguistic form. The source and target do not need identical syntax, but they must support the same conclusions and must not suggest a new mission action.

The diagnostic receipt for case 36 is small: retain the source phrase or field, record the rejected interpretation, state the revised wording, and indicate which source owner can resolve uncertainty. Then ask a second reader to explain the target without seeing the source. If they cannot distinguish prediction from observation, assessment from approval, or data availability from physical status where relevant, the revision is not finished. A public FAQ or search snippet derived from this case needs the same factual discipline as the full article.

Worked case 37: Uncertainty ellipse

Source fact: A diagram shows a conceptual uncertainty projection. Misleading target: Alt text calls the ellipse the actual physical location of a satellite. The issue here is not whether the target sounds natural, but which independent fact it causes the reader to believe. A translator should first identify the value, the source of authority, the time reference and the degree of certainty before selecting a natural sentence. If any of those properties are missing from the provided record, the draft should disclose the absence rather than filling it with a confident guess.

Defensible rendering: Describe a modeled uncertainty region, not a photographed path. Why the repair works: The visual encodes probability assumptions, not a direct picture. Treat this as a time-line problem. Distinguish document creation, observation, prediction, review, decision and publication. A date can be syntactically correct while answering the wrong temporal question. This is a practical exercise in reconstructing the same evidence in a different linguistic form. The source and target do not need identical syntax, but they must support the same conclusions and must not suggest a new mission action.

The diagnostic receipt for case 37 is small: retain the source phrase or field, record the rejected interpretation, state the revised wording, and indicate which source owner can resolve uncertainty. Then ask a second reader to explain the target without seeing the source. If they cannot distinguish prediction from observation, assessment from approval, or data availability from physical status where relevant, the revision is not finished. A public FAQ or search snippet derived from this case needs the same factual discipline as the full article.

Worked case 38: Maneuver after TCA

Source fact: A timeline places a proposed response after the predicted event. Misleading target: A copy editor silently moves the time to before TCA. The issue here is not whether the target sounds natural, but which independent fact it causes the reader to believe. A translator should first identify the value, the source of authority, the time reference and the degree of certainty before selecting a natural sentence. If any of those properties are missing from the provided record, the draft should disclose the absence rather than filling it with a confident guess.

Defensible rendering: Flag the inconsistency and query the source; never invent the intended schedule. Why the repair works: Correcting an operational plan is outside the translator’s authority. Treat this as a machine-human boundary problem. A parser keyword or ID must stay exact; a narrative definition can be translated and clarified. A reviewer should inspect both representations rather than assuming that correct prose guarantees a valid data feed. This is a practical exercise in reconstructing the same evidence in a different linguistic form. The source and target do not need identical syntax, but they must support the same conclusions and must not suggest a new mission action.

The diagnostic receipt for case 38 is small: retain the source phrase or field, record the rejected interpretation, state the revised wording, and indicate which source owner can resolve uncertainty. Then ask a second reader to explain the target without seeing the source. If they cannot distinguish prediction from observation, assessment from approval, or data availability from physical status where relevant, the revision is not finished. A public FAQ or search snippet derived from this case needs the same factual discipline as the full article.

Worked case 39: Two events, same pair

Source fact: The same two objects appear in separate screened encounters. Misleading target: A document merges their TCAs into one rolling alert. The issue here is not whether the target sounds natural, but which independent fact it causes the reader to believe. A translator should first identify the value, the source of authority, the time reference and the degree of certainty before selecting a natural sentence. If any of those properties are missing from the provided record, the draft should disclose the absence rather than filling it with a confident guess.

Defensible rendering: Keep separate event identities and time windows. Why the repair works: A pair identifier alone may not uniquely identify an encounter. Treat this as a reader-action problem. Imagine a non-specialist receiving only the translated summary. Would they infer a crash, an approved maneuver or a guarantee that the source never supplied? If so, the language needs a narrower assertion. This is a practical exercise in reconstructing the same evidence in a different linguistic form. The source and target do not need identical syntax, but they must support the same conclusions and must not suggest a new mission action.

The diagnostic receipt for case 39 is small: retain the source phrase or field, record the rejected interpretation, state the revised wording, and indicate which source owner can resolve uncertainty. Then ask a second reader to explain the target without seeing the source. If they cannot distinguish prediction from observation, assessment from approval, or data availability from physical status where relevant, the revision is not finished. A public FAQ or search snippet derived from this case needs the same factual discipline as the full article.

Worked case 40: Misidentified debris

Source fact: An unknown tracked object is classified as ‘unidentified’ in source. Misleading target: Translated public copy calls it a defunct rocket body. The issue here is not whether the target sounds natural, but which independent fact it causes the reader to believe. A translator should first identify the value, the source of authority, the time reference and the degree of certainty before selecting a natural sentence. If any of those properties are missing from the provided record, the draft should disclose the absence rather than filling it with a confident guess.

Defensible rendering: Keep the unknown classification and explain limitations. Why the repair works: Uncertainty about identity should survive localization. Treat this as a correction-readiness problem. Record enough source data to repair the public explanation after a future message changes the event. A source snapshot and a dated release note are more useful than an unlabeled copied paragraph. This is a practical exercise in reconstructing the same evidence in a different linguistic form. The source and target do not need identical syntax, but they must support the same conclusions and must not suggest a new mission action.

The diagnostic receipt for case 40 is small: retain the source phrase or field, record the rejected interpretation, state the revised wording, and indicate which source owner can resolve uncertainty. Then ask a second reader to explain the target without seeing the source. If they cannot distinguish prediction from observation, assessment from approval, or data availability from physical status where relevant, the revision is not finished. A public FAQ or search snippet derived from this case needs the same factual discipline as the full article.

Worked case 41: False current count

Source fact: An old space-debris population statistic is copied into a 2026 explainer as present-day data. Misleading target: The article makes a dynamic estimate sound timeless. The issue here is not whether the target sounds natural, but which independent fact it causes the reader to believe. A translator should first identify the value, the source of authority, the time reference and the degree of certainty before selecting a natural sentence. If any of those properties are missing from the provided record, the draft should disclose the absence rather than filling it with a confident guess.

Defensible rendering: Date every statistic and cite a reporting period. Why the repair works: Contextual numbers should be refreshed and distinguished from live surveillance. Treat this as a role-and-reference problem. Identify every noun phrase that points to a real object, a model state, a report or a decision. The target must preserve those referents even if word order changes across languages. This is a practical exercise in reconstructing the same evidence in a different linguistic form. The source and target do not need identical syntax, but they must support the same conclusions and must not suggest a new mission action.

The diagnostic receipt for case 41 is small: retain the source phrase or field, record the rejected interpretation, state the revised wording, and indicate which source owner can resolve uncertainty. Then ask a second reader to explain the target without seeing the source. If they cannot distinguish prediction from observation, assessment from approval, or data availability from physical status where relevant, the revision is not finished. A public FAQ or search snippet derived from this case needs the same factual discipline as the full article.

Worked case 42: ESA report timing

Source fact: A September 2026 environment report summarizes data collected through the end of 2025. Misleading target: A sentence says the 2026 report measured all conditions through October 2026. The issue here is not whether the target sounds natural, but which independent fact it causes the reader to believe. A translator should first identify the value, the source of authority, the time reference and the degree of certainty before selecting a natural sentence. If any of those properties are missing from the provided record, the draft should disclose the absence rather than filling it with a confident guess.

Defensible rendering: Give publication and observation cutoff separately. Why the repair works: A report year does not automatically equal a complete observation year. Treat this as an epistemic-strength problem. Write down whether the source asserts an observation, an estimate, a prediction, a judgment or a completed act. The target must not silently move the claim to a stronger category. This is a practical exercise in reconstructing the same evidence in a different linguistic form. The source and target do not need identical syntax, but they must support the same conclusions and must not suggest a new mission action.

The diagnostic receipt for case 42 is small: retain the source phrase or field, record the rejected interpretation, state the revised wording, and indicate which source owner can resolve uncertainty. Then ask a second reader to explain the target without seeing the source. If they cannot distinguish prediction from observation, assessment from approval, or data availability from physical status where relevant, the revision is not finished. A public FAQ or search snippet derived from this case needs the same factual discipline as the full article.

Worked case 43: Crossing into reentry

Source fact: An article on on-orbit conjunction includes a paragraph about atmospheric reentry casualty risk. Misleading target: The translation treats the two metrics as the same hazard probability. The issue here is not whether the target sounds natural, but which independent fact it causes the reader to believe. A translator should first identify the value, the source of authority, the time reference and the degree of certainty before selecting a natural sentence. If any of those properties are missing from the provided record, the draft should disclose the absence rather than filling it with a confident guess.

Defensible rendering: Explain the distinct domains and methods. Why the repair works: Two space-safety topics can be related without sharing the same event measure. Treat this as a numerical-invariant problem. Place the original token beside the rendered value, preserve any sign, unit, epoch or exponent, and run a back-conversion where applicable. Nice formatting cannot compensate for changed mathematics. This is a practical exercise in reconstructing the same evidence in a different linguistic form. The source and target do not need identical syntax, but they must support the same conclusions and must not suggest a new mission action.

The diagnostic receipt for case 43 is small: retain the source phrase or field, record the rejected interpretation, state the revised wording, and indicate which source owner can resolve uncertainty. Then ask a second reader to explain the target without seeing the source. If they cannot distinguish prediction from observation, assessment from approval, or data availability from physical status where relevant, the revision is not finished. A public FAQ or search snippet derived from this case needs the same factual discipline as the full article.

Worked case 44: Archived event summary

Source fact: A historian writes about an alert issued before a subsequent successful avoidance action. Misleading target: The target language collapses the full sequence into a timeless present. The issue here is not whether the target sounds natural, but which independent fact it causes the reader to believe. A translator should first identify the value, the source of authority, the time reference and the degree of certainty before selecting a natural sentence. If any of those properties are missing from the provided record, the draft should disclose the absence rather than filling it with a confident guess.

Defensible rendering: Retain chronology and source qualification for the outcome. Why the repair works: Historic accounts need event-time and knowledge-time precision. Treat this as a provenance problem. Identify who issued the record, which version supplied the fact and how much of the statement comes from that source. Never assemble one authoritative-sounding sentence out of incompatible snapshots. This is a practical exercise in reconstructing the same evidence in a different linguistic form. The source and target do not need identical syntax, but they must support the same conclusions and must not suggest a new mission action.

The diagnostic receipt for case 44 is small: retain the source phrase or field, record the rejected interpretation, state the revised wording, and indicate which source owner can resolve uncertainty. Then ask a second reader to explain the target without seeing the source. If they cannot distinguish prediction from observation, assessment from approval, or data availability from physical status where relevant, the revision is not finished. A public FAQ or search snippet derived from this case needs the same factual discipline as the full article.

Worked case 45: Public press release

Source fact: A mission team drafts an accessible update about a screened event awaiting review. Misleading target: Marketing language says experts ‘have prevented a collision’. The issue here is not whether the target sounds natural, but which independent fact it causes the reader to believe. A translator should first identify the value, the source of authority, the time reference and the degree of certainty before selecting a natural sentence. If any of those properties are missing from the provided record, the draft should disclose the absence rather than filling it with a confident guess.

Defensible rendering: State what is confirmed and avoid claiming successful prevention before evidence exists. Why the repair works: Public reassurance should not alter causal claims. Treat this as a time-line problem. Distinguish document creation, observation, prediction, review, decision and publication. A date can be syntactically correct while answering the wrong temporal question. This is a practical exercise in reconstructing the same evidence in a different linguistic form. The source and target do not need identical syntax, but they must support the same conclusions and must not suggest a new mission action.

The diagnostic receipt for case 45 is small: retain the source phrase or field, record the rejected interpretation, state the revised wording, and indicate which source owner can resolve uncertainty. Then ask a second reader to explain the target without seeing the source. If they cannot distinguish prediction from observation, assessment from approval, or data availability from physical status where relevant, the revision is not finished. A public FAQ or search snippet derived from this case needs the same factual discipline as the full article.

Worked case 46: Machine translation hallucination

Source fact: An AI-generated summary adds a probability not contained in the CDM. Misleading target: The reviewer assumes the value came from an invisible source. The issue here is not whether the target sounds natural, but which independent fact it causes the reader to believe. A translator should first identify the value, the source of authority, the time reference and the degree of certainty before selecting a natural sentence. If any of those properties are missing from the provided record, the draft should disclose the absence rather than filling it with a confident guess.

Defensible rendering: Delete invented numbers and record missing source evidence. Why the repair works: Language tools are not substitutes for mission data acquisition. Treat this as a machine-human boundary problem. A parser keyword or ID must stay exact; a narrative definition can be translated and clarified. A reviewer should inspect both representations rather than assuming that correct prose guarantees a valid data feed. This is a practical exercise in reconstructing the same evidence in a different linguistic form. The source and target do not need identical syntax, but they must support the same conclusions and must not suggest a new mission action.

The diagnostic receipt for case 46 is small: retain the source phrase or field, record the rejected interpretation, state the revised wording, and indicate which source owner can resolve uncertainty. Then ask a second reader to explain the target without seeing the source. If they cannot distinguish prediction from observation, assessment from approval, or data availability from physical status where relevant, the revision is not finished. A public FAQ or search snippet derived from this case needs the same factual discipline as the full article.

Worked case 47: Translation memory bleed

Source fact: A reused segment contains a prior satellite’s catalog number. Misleading target: The new alert inherits that object ID. The issue here is not whether the target sounds natural, but which independent fact it causes the reader to believe. A translator should first identify the value, the source of authority, the time reference and the degree of certainty before selecting a natural sentence. If any of those properties are missing from the provided record, the draft should disclose the absence rather than filling it with a confident guess.

Defensible rendering: Replace with the verified source value and strengthen variable locking. Why the repair works: Translation memory must not transfer event-specific identity. Treat this as a reader-action problem. Imagine a non-specialist receiving only the translated summary. Would they infer a crash, an approved maneuver or a guarantee that the source never supplied? If so, the language needs a narrower assertion. This is a practical exercise in reconstructing the same evidence in a different linguistic form. The source and target do not need identical syntax, but they must support the same conclusions and must not suggest a new mission action.

The diagnostic receipt for case 47 is small: retain the source phrase or field, record the rejected interpretation, state the revised wording, and indicate which source owner can resolve uncertainty. Then ask a second reader to explain the target without seeing the source. If they cannot distinguish prediction from observation, assessment from approval, or data availability from physical status where relevant, the revision is not finished. A public FAQ or search snippet derived from this case needs the same factual discipline as the full article.

Worked case 48: Glossary drift

Source fact: A termbase maps ‘conjunction’ to an ordinary astronomical alignment without qualification. Misleading target: A safety notice becomes a discussion of a planetary alignment. The issue here is not whether the target sounds natural, but which independent fact it causes the reader to believe. A translator should first identify the value, the source of authority, the time reference and the degree of certainty before selecting a natural sentence. If any of those properties are missing from the provided record, the draft should disclose the absence rather than filling it with a confident guess.

Defensible rendering: Define the orbital screening context and preserve the close-approach sense. Why the repair works: The correct equivalent depends on domain, not just dictionary frequency. Treat this as a correction-readiness problem. Record enough source data to repair the public explanation after a future message changes the event. A source snapshot and a dated release note are more useful than an unlabeled copied paragraph. This is a practical exercise in reconstructing the same evidence in a different linguistic form. The source and target do not need identical syntax, but they must support the same conclusions and must not suggest a new mission action.

The diagnostic receipt for case 48 is small: retain the source phrase or field, record the rejected interpretation, state the revised wording, and indicate which source owner can resolve uncertainty. Then ask a second reader to explain the target without seeing the source. If they cannot distinguish prediction from observation, assessment from approval, or data availability from physical status where relevant, the revision is not finished. A public FAQ or search snippet derived from this case needs the same factual discipline as the full article.

Worked case 49: Time zone display

Source fact: An English public page presents UTC and another language version shows a local time. Misleading target: The localized page does not say which time standard it uses. The issue here is not whether the target sounds natural, but which independent fact it causes the reader to believe. A translator should first identify the value, the source of authority, the time reference and the degree of certainty before selecting a natural sentence. If any of those properties are missing from the provided record, the draft should disclose the absence rather than filling it with a confident guess.

Defensible rendering: Label every clock, keep canonical UTC and test date rollover. Why the repair works: Multilingual presentation may differ, but underlying event instant must remain invariant. Treat this as a role-and-reference problem. Identify every noun phrase that points to a real object, a model state, a report or a decision. The target must preserve those referents even if word order changes across languages. This is a practical exercise in reconstructing the same evidence in a different linguistic form. The source and target do not need identical syntax, but they must support the same conclusions and must not suggest a new mission action.

The diagnostic receipt for case 49 is small: retain the source phrase or field, record the rejected interpretation, state the revised wording, and indicate which source owner can resolve uncertainty. Then ask a second reader to explain the target without seeing the source. If they cannot distinguish prediction from observation, assessment from approval, or data availability from physical status where relevant, the revision is not finished. A public FAQ or search snippet derived from this case needs the same factual discipline as the full article.

Worked case 50: International designation

Source fact: A launch-derived international designator and a catalog number appear in the same record. Misleading target: The target copy swaps their labels. The issue here is not whether the target sounds natural, but which independent fact it causes the reader to believe. A translator should first identify the value, the source of authority, the time reference and the degree of certainty before selecting a natural sentence. If any of those properties are missing from the provided record, the draft should disclose the absence rather than filling it with a confident guess.

Defensible rendering: Keep separate names and validate each against the originating record. Why the repair works: Identifiers may correlate without being interchangeable. Treat this as an epistemic-strength problem. Write down whether the source asserts an observation, an estimate, a prediction, a judgment or a completed act. The target must not silently move the claim to a stronger category. This is a practical exercise in reconstructing the same evidence in a different linguistic form. The source and target do not need identical syntax, but they must support the same conclusions and must not suggest a new mission action.

The diagnostic receipt for case 50 is small: retain the source phrase or field, record the rejected interpretation, state the revised wording, and indicate which source owner can resolve uncertainty. Then ask a second reader to explain the target without seeing the source. If they cannot distinguish prediction from observation, assessment from approval, or data availability from physical status where relevant, the revision is not finished. A public FAQ or search snippet derived from this case needs the same factual discipline as the full article.

Worked case 51: Data availability

Source fact: A requested object’s public feed is temporarily unavailable. Misleading target: The translation describes the object as lost or destroyed. The issue here is not whether the target sounds natural, but which independent fact it causes the reader to believe. A translator should first identify the value, the source of authority, the time reference and the degree of certainty before selecting a natural sentence. If any of those properties are missing from the provided record, the draft should disclose the absence rather than filling it with a confident guess.

Defensible rendering: State only that this data source has not returned a current record. Why the repair works: Information-system availability is not spacecraft physical status. Treat this as a numerical-invariant problem. Place the original token beside the rendered value, preserve any sign, unit, epoch or exponent, and run a back-conversion where applicable. Nice formatting cannot compensate for changed mathematics. This is a practical exercise in reconstructing the same evidence in a different linguistic form. The source and target do not need identical syntax, but they must support the same conclusions and must not suggest a new mission action.

The diagnostic receipt for case 51 is small: retain the source phrase or field, record the rejected interpretation, state the revised wording, and indicate which source owner can resolve uncertainty. Then ask a second reader to explain the target without seeing the source. If they cannot distinguish prediction from observation, assessment from approval, or data availability from physical status where relevant, the revision is not finished. A public FAQ or search snippet derived from this case needs the same factual discipline as the full article.

Worked case 52: Unit conversion

Source fact: A public explanation converts km to m. Misleading target: A value of 0.5 km is printed as 50 m. The issue here is not whether the target sounds natural, but which independent fact it causes the reader to believe. A translator should first identify the value, the source of authority, the time reference and the degree of certainty before selecting a natural sentence. If any of those properties are missing from the provided record, the draft should disclose the absence rather than filling it with a confident guess.

Defensible rendering: Check that 0.5 km equals 500 m and label the conversion. Why the repair works: Conversions should be reproducible and optional rather than hidden. Treat this as a provenance problem. Identify who issued the record, which version supplied the fact and how much of the statement comes from that source. Never assemble one authoritative-sounding sentence out of incompatible snapshots. This is a practical exercise in reconstructing the same evidence in a different linguistic form. The source and target do not need identical syntax, but they must support the same conclusions and must not suggest a new mission action.

The diagnostic receipt for case 52 is small: retain the source phrase or field, record the rejected interpretation, state the revised wording, and indicate which source owner can resolve uncertainty. Then ask a second reader to explain the target without seeing the source. If they cannot distinguish prediction from observation, assessment from approval, or data availability from physical status where relevant, the revision is not finished. A public FAQ or search snippet derived from this case needs the same factual discipline as the full article.

Worked case 53: Scientific decimal separator

Source fact: The source says 0.00002 and a localized layout prints 0,00002. Misleading target: A machine importer reads the comma as a delimiter. The issue here is not whether the target sounds natural, but which independent fact it causes the reader to believe. A translator should first identify the value, the source of authority, the time reference and the degree of certainty before selecting a natural sentence. If any of those properties are missing from the provided record, the draft should disclose the absence rather than filling it with a confident guess.

Defensible rendering: Keep machine format and human display channels distinct. Why the repair works: A language convention may be right for readers and wrong for parsers. Treat this as a time-line problem. Distinguish document creation, observation, prediction, review, decision and publication. A date can be syntactically correct while answering the wrong temporal question. This is a practical exercise in reconstructing the same evidence in a different linguistic form. The source and target do not need identical syntax, but they must support the same conclusions and must not suggest a new mission action.

The diagnostic receipt for case 53 is small: retain the source phrase or field, record the rejected interpretation, state the revised wording, and indicate which source owner can resolve uncertainty. Then ask a second reader to explain the target without seeing the source. If they cannot distinguish prediction from observation, assessment from approval, or data availability from physical status where relevant, the revision is not finished. A public FAQ or search snippet derived from this case needs the same factual discipline as the full article.

Worked case 54: Approval date

Source fact: An advisory is dated Monday but flight review occurred on Tuesday. Misleading target: A multilingual version says the decision was made on Monday. The issue here is not whether the target sounds natural, but which independent fact it causes the reader to believe. A translator should first identify the value, the source of authority, the time reference and the degree of certainty before selecting a natural sentence. If any of those properties are missing from the provided record, the draft should disclose the absence rather than filling it with a confident guess.

Defensible rendering: List issuer date and decision date separately. Why the repair works: Authority-related chronology cannot be reconstructed from one headline. Treat this as a machine-human boundary problem. A parser keyword or ID must stay exact; a narrative definition can be translated and clarified. A reviewer should inspect both representations rather than assuming that correct prose guarantees a valid data feed. This is a practical exercise in reconstructing the same evidence in a different linguistic form. The source and target do not need identical syntax, but they must support the same conclusions and must not suggest a new mission action.

The diagnostic receipt for case 54 is small: retain the source phrase or field, record the rejected interpretation, state the revised wording, and indicate which source owner can resolve uncertainty. Then ask a second reader to explain the target without seeing the source. If they cannot distinguish prediction from observation, assessment from approval, or data availability from physical status where relevant, the revision is not finished. A public FAQ or search snippet derived from this case needs the same factual discipline as the full article.

Worked case 55: Quotation boundaries

Source fact: A report quotes a provider warning inside commentary. Misleading target: The translation removes quotation marks and attributes the warning to its own author. The issue here is not whether the target sounds natural, but which independent fact it causes the reader to believe. A translator should first identify the value, the source of authority, the time reference and the degree of certainty before selecting a natural sentence. If any of those properties are missing from the provided record, the draft should disclose the absence rather than filling it with a confident guess.

Defensible rendering: Keep speech attribution and quoted status visible. Why the repair works: Readers should know who made the assertion. Treat this as a reader-action problem. Imagine a non-specialist receiving only the translated summary. Would they infer a crash, an approved maneuver or a guarantee that the source never supplied? If so, the language needs a narrower assertion. This is a practical exercise in reconstructing the same evidence in a different linguistic form. The source and target do not need identical syntax, but they must support the same conclusions and must not suggest a new mission action.

The diagnostic receipt for case 55 is small: retain the source phrase or field, record the rejected interpretation, state the revised wording, and indicate which source owner can resolve uncertainty. Then ask a second reader to explain the target without seeing the source. If they cannot distinguish prediction from observation, assessment from approval, or data availability from physical status where relevant, the revision is not finished. A public FAQ or search snippet derived from this case needs the same factual discipline as the full article.

Worked case 56: Status reversal

Source fact: An event was reopened after an earlier closeout. Misleading target: The target page keeps the previous ‘closed’ badge without qualification. The issue here is not whether the target sounds natural, but which independent fact it causes the reader to believe. A translator should first identify the value, the source of authority, the time reference and the degree of certainty before selecting a natural sentence. If any of those properties are missing from the provided record, the draft should disclose the absence rather than filling it with a confident guess.

Defensible rendering: Explain the reopening and link to the new revision. Why the repair works: Lifecycle status can reverse as new data arrive. Treat this as a correction-readiness problem. Record enough source data to repair the public explanation after a future message changes the event. A source snapshot and a dated release note are more useful than an unlabeled copied paragraph. This is a practical exercise in reconstructing the same evidence in a different linguistic form. The source and target do not need identical syntax, but they must support the same conclusions and must not suggest a new mission action.

The diagnostic receipt for case 56 is small: retain the source phrase or field, record the rejected interpretation, state the revised wording, and indicate which source owner can resolve uncertainty. Then ask a second reader to explain the target without seeing the source. If they cannot distinguish prediction from observation, assessment from approval, or data availability from physical status where relevant, the revision is not finished. A public FAQ or search snippet derived from this case needs the same factual discipline as the full article.

Worked case 57: Unreviewed animation

Source fact: An orbital animation visually depicts a near miss using simulated motion. Misleading target: A caption calls it a camera recording. The issue here is not whether the target sounds natural, but which independent fact it causes the reader to believe. A translator should first identify the value, the source of authority, the time reference and the degree of certainty before selecting a natural sentence. If any of those properties are missing from the provided record, the draft should disclose the absence rather than filling it with a confident guess.

Defensible rendering: Label animation and underlying scenario model. Why the repair works: Illustrations must not be promoted into direct evidence. Treat this as a role-and-reference problem. Identify every noun phrase that points to a real object, a model state, a report or a decision. The target must preserve those referents even if word order changes across languages. This is a practical exercise in reconstructing the same evidence in a different linguistic form. The source and target do not need identical syntax, but they must support the same conclusions and must not suggest a new mission action.

The diagnostic receipt for case 57 is small: retain the source phrase or field, record the rejected interpretation, state the revised wording, and indicate which source owner can resolve uncertainty. Then ask a second reader to explain the target without seeing the source. If they cannot distinguish prediction from observation, assessment from approval, or data availability from physical status where relevant, the revision is not finished. A public FAQ or search snippet derived from this case needs the same factual discipline as the full article.

Worked case 58: Public-facing FAQ

Source fact: A reader asks ‘Is a conjunction always dangerous?’ Misleading target: The translated answer says all conjunctions require maneuvers. The issue here is not whether the target sounds natural, but which independent fact it causes the reader to believe. A translator should first identify the value, the source of authority, the time reference and the degree of certainty before selecting a natural sentence. If any of those properties are missing from the provided record, the draft should disclose the absence rather than filling it with a confident guess.

Defensible rendering: Explain that screening identifies candidates for assessment, not automatic burn decisions. Why the repair works: Meaningful public education distinguishes detection from authorized action. Treat this as an epistemic-strength problem. Write down whether the source asserts an observation, an estimate, a prediction, a judgment or a completed act. The target must not silently move the claim to a stronger category. This is a practical exercise in reconstructing the same evidence in a different linguistic form. The source and target do not need identical syntax, but they must support the same conclusions and must not suggest a new mission action.

The diagnostic receipt for case 58 is small: retain the source phrase or field, record the rejected interpretation, state the revised wording, and indicate which source owner can resolve uncertainty. Then ask a second reader to explain the target without seeing the source. If they cannot distinguish prediction from observation, assessment from approval, or data availability from physical status where relevant, the revision is not finished. A public FAQ or search snippet derived from this case needs the same factual discipline as the full article.

Worked case 59: Incident correction

Source fact: A statement about TCA was mistranslated in an earlier release. Misleading target: The correction silently replaces text without note. The issue here is not whether the target sounds natural, but which independent fact it causes the reader to believe. A translator should first identify the value, the source of authority, the time reference and the degree of certainty before selecting a natural sentence. If any of those properties are missing from the provided record, the draft should disclose the absence rather than filling it with a confident guess.

Defensible rendering: Add a dated correction naming the changed claim and source record. Why the repair works: Trust depends on traceable repair, not pretending a mistake never happened. Treat this as a numerical-invariant problem. Place the original token beside the rendered value, preserve any sign, unit, epoch or exponent, and run a back-conversion where applicable. Nice formatting cannot compensate for changed mathematics. This is a practical exercise in reconstructing the same evidence in a different linguistic form. The source and target do not need identical syntax, but they must support the same conclusions and must not suggest a new mission action.

The diagnostic receipt for case 59 is small: retain the source phrase or field, record the rejected interpretation, state the revised wording, and indicate which source owner can resolve uncertainty. Then ask a second reader to explain the target without seeing the source. If they cannot distinguish prediction from observation, assessment from approval, or data availability from physical status where relevant, the revision is not finished. A public FAQ or search snippet derived from this case needs the same factual discipline as the full article.

Worked case 60: Cross-lingual accessibility

Source fact: A screen reader reads TCA as one unfamiliar word and skips explanation. Misleading target: The public page assumes acronyms are universally understood. The issue here is not whether the target sounds natural, but which independent fact it causes the reader to believe. A translator should first identify the value, the source of authority, the time reference and the degree of certainty before selecting a natural sentence. If any of those properties are missing from the provided record, the draft should disclose the absence rather than filling it with a confident guess.

Defensible rendering: Expand TCA once as ‘time of closest approach’ and retain the abbreviation. Why the repair works: Accessibility requires concept explanation without changing semantics. Treat this as a provenance problem. Identify who issued the record, which version supplied the fact and how much of the statement comes from that source. Never assemble one authoritative-sounding sentence out of incompatible snapshots. This is a practical exercise in reconstructing the same evidence in a different linguistic form. The source and target do not need identical syntax, but they must support the same conclusions and must not suggest a new mission action.

The diagnostic receipt for case 60 is small: retain the source phrase or field, record the rejected interpretation, state the revised wording, and indicate which source owner can resolve uncertainty. Then ask a second reader to explain the target without seeing the source. If they cannot distinguish prediction from observation, assessment from approval, or data availability from physical status where relevant, the revision is not finished. A public FAQ or search snippet derived from this case needs the same factual discipline as the full article.

Worked case 61: Third-party repost

Source fact: A news site repeats a translated alert after the original source revises it. Misleading target: The repost presents its earlier account as live information. The issue here is not whether the target sounds natural, but which independent fact it causes the reader to believe. A translator should first identify the value, the source of authority, the time reference and the degree of certainty before selecting a natural sentence. If any of those properties are missing from the provided record, the draft should disclose the absence rather than filling it with a confident guess.

Defensible rendering: Date, source and status the excerpt and route to the authoritative update. Why the repair works: Publication chains have separate correction obligations. Treat this as a time-line problem. Distinguish document creation, observation, prediction, review, decision and publication. A date can be syntactically correct while answering the wrong temporal question. This is a practical exercise in reconstructing the same evidence in a different linguistic form. The source and target do not need identical syntax, but they must support the same conclusions and must not suggest a new mission action.

The diagnostic receipt for case 61 is small: retain the source phrase or field, record the rejected interpretation, state the revised wording, and indicate which source owner can resolve uncertainty. Then ask a second reader to explain the target without seeing the source. If they cannot distinguish prediction from observation, assessment from approval, or data availability from physical status where relevant, the revision is not finished. A public FAQ or search snippet derived from this case needs the same factual discipline as the full article.

Worked case 62: Operational disclaimer

Source fact: An educational example includes fictional values similar to real CDM fields. Misleading target: Readers mistake it for an actual operational collision alert. The issue here is not whether the target sounds natural, but which independent fact it causes the reader to believe. A translator should first identify the value, the source of authority, the time reference and the degree of certainty before selecting a natural sentence. If any of those properties are missing from the provided record, the draft should disclose the absence rather than filling it with a confident guess.

Defensible rendering: Label all figures fictional, without real decisions or maneuver instructions. Why the repair works: A teaching document should not masquerade as live safety information. Treat this as a machine-human boundary problem. A parser keyword or ID must stay exact; a narrative definition can be translated and clarified. A reviewer should inspect both representations rather than assuming that correct prose guarantees a valid data feed. This is a practical exercise in reconstructing the same evidence in a different linguistic form. The source and target do not need identical syntax, but they must support the same conclusions and must not suggest a new mission action.

The diagnostic receipt for case 62 is small: retain the source phrase or field, record the rejected interpretation, state the revised wording, and indicate which source owner can resolve uncertainty. Then ask a second reader to explain the target without seeing the source. If they cannot distinguish prediction from observation, assessment from approval, or data availability from physical status where relevant, the revision is not finished. A public FAQ or search snippet derived from this case needs the same factual discipline as the full article.

Transfer workshops: use the same thinking beyond orbital alerts

The specialist skill here is transferable: identify the entity, evidence, time, model, uncertainty, authority, action and revision. Apply the method to other domains without importing the spacecraft-specific definitions into unrelated systems. Each workshop describes a related meaning-preservation problem and the boundary that makes its solution distinct.

Workshop 1: Air traffic conflict alerts

A predicted aircraft conflict is not the same as an actual collision; keep time, relative geometry and decision authority separate. Explain which surveillance system and procedure produced the alert rather than copying the satellite risk threshold. To practise, take a short source statement and produce three outputs: a technically literal inventory of facts, a natural-language explanation that preserves every qualifier, and a one-sentence headline that cannot be mistaken for a stronger claim. Compare the three outputs against the original, highlight any missing actor or time scope, and return uncertain information to its rightful owner instead of inferring it. This workshop transfers a reasoning method, not an orbital safety rule.

The pass criterion is that an independent reader can explain what is known, what is inferred, who owns the judgment and what would count as a later correction. If the target language encourages a categorical sentence where the source is explicitly conditional, rewrite it with a clear qualifier. For air traffic conflict alerts, the technical vocabulary is domain-specific; build a new termbase rather than copying spacecraft terms into a new field.

Workshop 2: Maritime near-miss reports

A closest-point-of-approach estimate is conditional on available motion data and prediction assumptions. Keep estimated passing distance distinct from an observed contact or confirmed damage. To practise, take a short source statement and produce three outputs: a technically literal inventory of facts, a natural-language explanation that preserves every qualifier, and a one-sentence headline that cannot be mistaken for a stronger claim. Compare the three outputs against the original, highlight any missing actor or time scope, and return uncertain information to its rightful owner instead of inferring it. This workshop transfers a reasoning method, not an orbital safety rule.

The pass criterion is that an independent reader can explain what is known, what is inferred, who owns the judgment and what would count as a later correction. If the target language encourages a categorical sentence where the source is explicitly conditional, rewrite it with a clear qualifier. For maritime near-miss reports, the technical vocabulary is domain-specific; build a new termbase rather than copying spacecraft terms into a new field.

Workshop 3: Earthquake early warning

A forecast or alert issued from sensor analysis is not a confirmation that every locality will experience a particular intensity. Protect time, uncertainty, spatial coverage and authority in public language. To practise, take a short source statement and produce three outputs: a technically literal inventory of facts, a natural-language explanation that preserves every qualifier, and a one-sentence headline that cannot be mistaken for a stronger claim. Compare the three outputs against the original, highlight any missing actor or time scope, and return uncertain information to its rightful owner instead of inferring it. This workshop transfers a reasoning method, not an orbital safety rule.

The pass criterion is that an independent reader can explain what is known, what is inferred, who owns the judgment and what would count as a later correction. If the target language encourages a categorical sentence where the source is explicitly conditional, rewrite it with a clear qualifier. For earthquake early warning, the technical vocabulary is domain-specific; build a new termbase rather than copying spacecraft terms into a new field.

Workshop 4: Weather model guidance

A forecast track and probability cone are not direct observations or promises of exact conditions. State model horizon, forecast issue and uncertainty where supplied. To practise, take a short source statement and produce three outputs: a technically literal inventory of facts, a natural-language explanation that preserves every qualifier, and a one-sentence headline that cannot be mistaken for a stronger claim. Compare the three outputs against the original, highlight any missing actor or time scope, and return uncertain information to its rightful owner instead of inferring it. This workshop transfers a reasoning method, not an orbital safety rule.

The pass criterion is that an independent reader can explain what is known, what is inferred, who owns the judgment and what would count as a later correction. If the target language encourages a categorical sentence where the source is explicitly conditional, rewrite it with a clear qualifier. For weather model guidance, the technical vocabulary is domain-specific; build a new termbase rather than copying spacecraft terms into a new field.

Workshop 5: Medical risk communication

A modeled probability in a defined cohort is not the individual’s guaranteed outcome. Retain denominator, applicability, date and uncertainty. To practise, take a short source statement and produce three outputs: a technically literal inventory of facts, a natural-language explanation that preserves every qualifier, and a one-sentence headline that cannot be mistaken for a stronger claim. Compare the three outputs against the original, highlight any missing actor or time scope, and return uncertain information to its rightful owner instead of inferring it. This workshop transfers a reasoning method, not an orbital safety rule.

The pass criterion is that an independent reader can explain what is known, what is inferred, who owns the judgment and what would count as a later correction. If the target language encourages a categorical sentence where the source is explicitly conditional, rewrite it with a clear qualifier. For medical risk communication, the technical vocabulary is domain-specific; build a new termbase rather than copying spacecraft terms into a new field.

Workshop 6: Epidemiological surveillance

Cases reported in a time window differ from real-time infections and modelled prevalence. Keep report date, reference period and case definition. To practise, take a short source statement and produce three outputs: a technically literal inventory of facts, a natural-language explanation that preserves every qualifier, and a one-sentence headline that cannot be mistaken for a stronger claim. Compare the three outputs against the original, highlight any missing actor or time scope, and return uncertain information to its rightful owner instead of inferring it. This workshop transfers a reasoning method, not an orbital safety rule.

The pass criterion is that an independent reader can explain what is known, what is inferred, who owns the judgment and what would count as a later correction. If the target language encourages a categorical sentence where the source is explicitly conditional, rewrite it with a clear qualifier. For epidemiological surveillance, the technical vocabulary is domain-specific; build a new termbase rather than copying spacecraft terms into a new field.

Workshop 7: Seismic catalogues

An initial event location can be revised following new evidence. Version estimated parameters rather than overwriting history without note. To practise, take a short source statement and produce three outputs: a technically literal inventory of facts, a natural-language explanation that preserves every qualifier, and a one-sentence headline that cannot be mistaken for a stronger claim. Compare the three outputs against the original, highlight any missing actor or time scope, and return uncertain information to its rightful owner instead of inferring it. This workshop transfers a reasoning method, not an orbital safety rule.

The pass criterion is that an independent reader can explain what is known, what is inferred, who owns the judgment and what would count as a later correction. If the target language encourages a categorical sentence where the source is explicitly conditional, rewrite it with a clear qualifier. For seismic catalogues, the technical vocabulary is domain-specific; build a new termbase rather than copying spacecraft terms into a new field.

Workshop 8: Fisheries stock models

A quota decision and a stock-assessment estimate are different authority and evidence stages. Keep policy judgment distinct from model-based biomass estimates. To practise, take a short source statement and produce three outputs: a technically literal inventory of facts, a natural-language explanation that preserves every qualifier, and a one-sentence headline that cannot be mistaken for a stronger claim. Compare the three outputs against the original, highlight any missing actor or time scope, and return uncertain information to its rightful owner instead of inferring it. This workshop transfers a reasoning method, not an orbital safety rule.

The pass criterion is that an independent reader can explain what is known, what is inferred, who owns the judgment and what would count as a later correction. If the target language encourages a categorical sentence where the source is explicitly conditional, rewrite it with a clear qualifier. For fisheries stock models, the technical vocabulary is domain-specific; build a new termbase rather than copying spacecraft terms into a new field.

Workshop 9: Ocean float observations

Raw instrument output and adjusted quality-controlled data should not be merged. Publish flags and adjustment provenance beside the translated descriptions. To practise, take a short source statement and produce three outputs: a technically literal inventory of facts, a natural-language explanation that preserves every qualifier, and a one-sentence headline that cannot be mistaken for a stronger claim. Compare the three outputs against the original, highlight any missing actor or time scope, and return uncertain information to its rightful owner instead of inferring it. This workshop transfers a reasoning method, not an orbital safety rule.

The pass criterion is that an independent reader can explain what is known, what is inferred, who owns the judgment and what would count as a later correction. If the target language encourages a categorical sentence where the source is explicitly conditional, rewrite it with a clear qualifier. For ocean float observations, the technical vocabulary is domain-specific; build a new termbase rather than copying spacecraft terms into a new field.

Workshop 10: Glacier mass balance

A measured winter balance and an annual modeled water-equivalent estimate have different scopes. Keep interval and unit even when terms are simplified for readers. To practise, take a short source statement and produce three outputs: a technically literal inventory of facts, a natural-language explanation that preserves every qualifier, and a one-sentence headline that cannot be mistaken for a stronger claim. Compare the three outputs against the original, highlight any missing actor or time scope, and return uncertain information to its rightful owner instead of inferring it. This workshop transfers a reasoning method, not an orbital safety rule.

The pass criterion is that an independent reader can explain what is known, what is inferred, who owns the judgment and what would count as a later correction. If the target language encourages a categorical sentence where the source is explicitly conditional, rewrite it with a clear qualifier. For glacier mass balance, the technical vocabulary is domain-specific; build a new termbase rather than copying spacecraft terms into a new field.

Workshop 11: Soil classification

A field description, laboratory test and classification label occupy distinct evidential levels. Do not translate a tentative taxonomy as a definitive observed fact. To practise, take a short source statement and produce three outputs: a technically literal inventory of facts, a natural-language explanation that preserves every qualifier, and a one-sentence headline that cannot be mistaken for a stronger claim. Compare the three outputs against the original, highlight any missing actor or time scope, and return uncertain information to its rightful owner instead of inferring it. This workshop transfers a reasoning method, not an orbital safety rule.

The pass criterion is that an independent reader can explain what is known, what is inferred, who owns the judgment and what would count as a later correction. If the target language encourages a categorical sentence where the source is explicitly conditional, rewrite it with a clear qualifier. For soil classification, the technical vocabulary is domain-specific; build a new termbase rather than copying spacecraft terms into a new field.

Workshop 12: Museum collection records

A historical attribution is not the same as proof of original authorship. Retain voice, evidence and provenance in every language. To practise, take a short source statement and produce three outputs: a technically literal inventory of facts, a natural-language explanation that preserves every qualifier, and a one-sentence headline that cannot be mistaken for a stronger claim. Compare the three outputs against the original, highlight any missing actor or time scope, and return uncertain information to its rightful owner instead of inferring it. This workshop transfers a reasoning method, not an orbital safety rule.

The pass criterion is that an independent reader can explain what is known, what is inferred, who owns the judgment and what would count as a later correction. If the target language encourages a categorical sentence where the source is explicitly conditional, rewrite it with a clear qualifier. For museum collection records, the technical vocabulary is domain-specific; build a new termbase rather than copying spacecraft terms into a new field.

Workshop 13: Digital library metadata

A catalog record may reference a work, edition, manifestation or physical item. Keep identifier scope and do not merge separately owned objects. To practise, take a short source statement and produce three outputs: a technically literal inventory of facts, a natural-language explanation that preserves every qualifier, and a one-sentence headline that cannot be mistaken for a stronger claim. Compare the three outputs against the original, highlight any missing actor or time scope, and return uncertain information to its rightful owner instead of inferring it. This workshop transfers a reasoning method, not an orbital safety rule.

The pass criterion is that an independent reader can explain what is known, what is inferred, who owns the judgment and what would count as a later correction. If the target language encourages a categorical sentence where the source is explicitly conditional, rewrite it with a clear qualifier. For digital library metadata, the technical vocabulary is domain-specific; build a new termbase rather than copying spacecraft terms into a new field.

Workshop 14: Financial risk disclosure

Scenario analysis differs from realized loss and management action. Preserve conditions, accounting period, unit and authority. To practise, take a short source statement and produce three outputs: a technically literal inventory of facts, a natural-language explanation that preserves every qualifier, and a one-sentence headline that cannot be mistaken for a stronger claim. Compare the three outputs against the original, highlight any missing actor or time scope, and return uncertain information to its rightful owner instead of inferring it. This workshop transfers a reasoning method, not an orbital safety rule.

The pass criterion is that an independent reader can explain what is known, what is inferred, who owns the judgment and what would count as a later correction. If the target language encourages a categorical sentence where the source is explicitly conditional, rewrite it with a clear qualifier. For financial risk disclosure, the technical vocabulary is domain-specific; build a new termbase rather than copying spacecraft terms into a new field.

Workshop 15: Software incident alerts

A monitoring warning is not confirmation of a customer-visible outage. Keep detection, investigation, mitigation and resolved statuses. To practise, take a short source statement and produce three outputs: a technically literal inventory of facts, a natural-language explanation that preserves every qualifier, and a one-sentence headline that cannot be mistaken for a stronger claim. Compare the three outputs against the original, highlight any missing actor or time scope, and return uncertain information to its rightful owner instead of inferring it. This workshop transfers a reasoning method, not an orbital safety rule.

The pass criterion is that an independent reader can explain what is known, what is inferred, who owns the judgment and what would count as a later correction. If the target language encourages a categorical sentence where the source is explicitly conditional, rewrite it with a clear qualifier. For software incident alerts, the technical vocabulary is domain-specific; build a new termbase rather than copying spacecraft terms into a new field.

Workshop 16: Robotics obstacle detection

An obstacle candidate is not a confirmed collision and a suggested path is not an executed command. Explain the distinction while leaving machine control tokens untouched. To practise, take a short source statement and produce three outputs: a technically literal inventory of facts, a natural-language explanation that preserves every qualifier, and a one-sentence headline that cannot be mistaken for a stronger claim. Compare the three outputs against the original, highlight any missing actor or time scope, and return uncertain information to its rightful owner instead of inferring it. This workshop transfers a reasoning method, not an orbital safety rule.

The pass criterion is that an independent reader can explain what is known, what is inferred, who owns the judgment and what would count as a later correction. If the target language encourages a categorical sentence where the source is explicitly conditional, rewrite it with a clear qualifier. For robotics obstacle detection, the technical vocabulary is domain-specific; build a new termbase rather than copying spacecraft terms into a new field.

Workshop 17: Industrial maintenance

A condition-monitoring alert is not automatically a work order. Carry the approval chain and evidence source into the translation. To practise, take a short source statement and produce three outputs: a technically literal inventory of facts, a natural-language explanation that preserves every qualifier, and a one-sentence headline that cannot be mistaken for a stronger claim. Compare the three outputs against the original, highlight any missing actor or time scope, and return uncertain information to its rightful owner instead of inferring it. This workshop transfers a reasoning method, not an orbital safety rule.

The pass criterion is that an independent reader can explain what is known, what is inferred, who owns the judgment and what would count as a later correction. If the target language encourages a categorical sentence where the source is explicitly conditional, rewrite it with a clear qualifier. For industrial maintenance, the technical vocabulary is domain-specific; build a new termbase rather than copying spacecraft terms into a new field.

Workshop 18: Satellite remote sensing

An inferred surface property is not necessarily a direct instrument measurement. Keep product level, processing method and spatial limitations. To practise, take a short source statement and produce three outputs: a technically literal inventory of facts, a natural-language explanation that preserves every qualifier, and a one-sentence headline that cannot be mistaken for a stronger claim. Compare the three outputs against the original, highlight any missing actor or time scope, and return uncertain information to its rightful owner instead of inferring it. This workshop transfers a reasoning method, not an orbital safety rule.

The pass criterion is that an independent reader can explain what is known, what is inferred, who owns the judgment and what would count as a later correction. If the target language encourages a categorical sentence where the source is explicitly conditional, rewrite it with a clear qualifier. For satellite remote sensing, the technical vocabulary is domain-specific; build a new termbase rather than copying spacecraft terms into a new field.

Workshop 19: Astronomical transient alerts

A candidate transient and a confirmed source identification are different claims. Retain follow-up status, coordinates, epochs and attribution. To practise, take a short source statement and produce three outputs: a technically literal inventory of facts, a natural-language explanation that preserves every qualifier, and a one-sentence headline that cannot be mistaken for a stronger claim. Compare the three outputs against the original, highlight any missing actor or time scope, and return uncertain information to its rightful owner instead of inferring it. This workshop transfers a reasoning method, not an orbital safety rule.

The pass criterion is that an independent reader can explain what is known, what is inferred, who owns the judgment and what would count as a later correction. If the target language encourages a categorical sentence where the source is explicitly conditional, rewrite it with a clear qualifier. For astronomical transient alerts, the technical vocabulary is domain-specific; build a new termbase rather than copying spacecraft terms into a new field.

Workshop 20: Election result reporting

A provisional count differs from a certified result. Keep the source, certification state and timestamp. To practise, take a short source statement and produce three outputs: a technically literal inventory of facts, a natural-language explanation that preserves every qualifier, and a one-sentence headline that cannot be mistaken for a stronger claim. Compare the three outputs against the original, highlight any missing actor or time scope, and return uncertain information to its rightful owner instead of inferring it. This workshop transfers a reasoning method, not an orbital safety rule.

The pass criterion is that an independent reader can explain what is known, what is inferred, who owns the judgment and what would count as a later correction. If the target language encourages a categorical sentence where the source is explicitly conditional, rewrite it with a clear qualifier. For election result reporting, the technical vocabulary is domain-specific; build a new termbase rather than copying spacecraft terms into a new field.

Workshop 21: Emergency public warnings

An active warning, updated warning and all-clear notice have different meanings. Track status and issuance authority across language versions. To practise, take a short source statement and produce three outputs: a technically literal inventory of facts, a natural-language explanation that preserves every qualifier, and a one-sentence headline that cannot be mistaken for a stronger claim. Compare the three outputs against the original, highlight any missing actor or time scope, and return uncertain information to its rightful owner instead of inferring it. This workshop transfers a reasoning method, not an orbital safety rule.

The pass criterion is that an independent reader can explain what is known, what is inferred, who owns the judgment and what would count as a later correction. If the target language encourages a categorical sentence where the source is explicitly conditional, rewrite it with a clear qualifier. For emergency public warnings, the technical vocabulary is domain-specific; build a new termbase rather than copying spacecraft terms into a new field.

Workshop 22: Accessibility annotations

A graph’s alt text must not remove uncertainty encoded by its legend. Describe the model quantity and scale, not an invented outcome. To practise, take a short source statement and produce three outputs: a technically literal inventory of facts, a natural-language explanation that preserves every qualifier, and a one-sentence headline that cannot be mistaken for a stronger claim. Compare the three outputs against the original, highlight any missing actor or time scope, and return uncertain information to its rightful owner instead of inferring it. This workshop transfers a reasoning method, not an orbital safety rule.

The pass criterion is that an independent reader can explain what is known, what is inferred, who owns the judgment and what would count as a later correction. If the target language encourages a categorical sentence where the source is explicitly conditional, rewrite it with a clear qualifier. For accessibility annotations, the technical vocabulary is domain-specific; build a new termbase rather than copying spacecraft terms into a new field.

Workshop 23: Legal drafting

A recommendation should not become a command or an enforceable mandate without authority. Treat modal verbs, scope and jurisdiction as structural meaning. To practise, take a short source statement and produce three outputs: a technically literal inventory of facts, a natural-language explanation that preserves every qualifier, and a one-sentence headline that cannot be mistaken for a stronger claim. Compare the three outputs against the original, highlight any missing actor or time scope, and return uncertain information to its rightful owner instead of inferring it. This workshop transfers a reasoning method, not an orbital safety rule.

The pass criterion is that an independent reader can explain what is known, what is inferred, who owns the judgment and what would count as a later correction. If the target language encourages a categorical sentence where the source is explicitly conditional, rewrite it with a clear qualifier. For legal drafting, the technical vocabulary is domain-specific; build a new termbase rather than copying spacecraft terms into a new field.

Workshop 24: Academic peer review

Preliminary findings differ from independently replicated conclusions. Keep study status, caveats and evidence source. To practise, take a short source statement and produce three outputs: a technically literal inventory of facts, a natural-language explanation that preserves every qualifier, and a one-sentence headline that cannot be mistaken for a stronger claim. Compare the three outputs against the original, highlight any missing actor or time scope, and return uncertain information to its rightful owner instead of inferring it. This workshop transfers a reasoning method, not an orbital safety rule.

The pass criterion is that an independent reader can explain what is known, what is inferred, who owns the judgment and what would count as a later correction. If the target language encourages a categorical sentence where the source is explicitly conditional, rewrite it with a clear qualifier. For academic peer review, the technical vocabulary is domain-specific; build a new termbase rather than copying spacecraft terms into a new field.

Workshop 25: Historical chronicles

The date an account was written differs from the date an event occurred. Preserve chronology and attribution. To practise, take a short source statement and produce three outputs: a technically literal inventory of facts, a natural-language explanation that preserves every qualifier, and a one-sentence headline that cannot be mistaken for a stronger claim. Compare the three outputs against the original, highlight any missing actor or time scope, and return uncertain information to its rightful owner instead of inferring it. This workshop transfers a reasoning method, not an orbital safety rule.

The pass criterion is that an independent reader can explain what is known, what is inferred, who owns the judgment and what would count as a later correction. If the target language encourages a categorical sentence where the source is explicitly conditional, rewrite it with a clear qualifier. For historical chronicles, the technical vocabulary is domain-specific; build a new termbase rather than copying spacecraft terms into a new field.

Workshop 26: AI-generated summaries

A system may invent confident details when condensing structured data. Verify each claim against raw fields before sharing it. To practise, take a short source statement and produce three outputs: a technically literal inventory of facts, a natural-language explanation that preserves every qualifier, and a one-sentence headline that cannot be mistaken for a stronger claim. Compare the three outputs against the original, highlight any missing actor or time scope, and return uncertain information to its rightful owner instead of inferring it. This workshop transfers a reasoning method, not an orbital safety rule.

The pass criterion is that an independent reader can explain what is known, what is inferred, who owns the judgment and what would count as a later correction. If the target language encourages a categorical sentence where the source is explicitly conditional, rewrite it with a clear qualifier. For ai-generated summaries, the technical vocabulary is domain-specific; build a new termbase rather than copying spacecraft terms into a new field.

Terminology: seventy-three quick definitions with usage boundaries

The following concise definitions are study aids, not replacements for the current domain standards or a mission-approved glossary. Where a technical term has multiple formal definitions, use the definition belonging to the applicable source document and its edition. In a multilingual production environment, store each concept with its reference and context before selecting target-language equivalents.

1. conjunction. A predicted close approach between tracked space objects in the relevant orbital-safety context. In translation, identify where the term appears and whether it is a data field, an analytic concept, a policy label or explanatory prose; only then select the target expression.

2. CDM. Conjunction Data Message; a standardized way to exchange relevant conjunction information. In translation, identify where the term appears and whether it is a data field, an analytic concept, a policy label or explanatory prose; only then select the target expression.

3. TCA. Time of closest approach; the modeled time of minimum separation for an event. In translation, identify where the term appears and whether it is a data field, an analytic concept, a policy label or explanatory prose; only then select the target expression.

4. miss distance. Predicted separation metric associated with the modeled encounter. In translation, identify where the term appears and whether it is a data field, an analytic concept, a policy label or explanatory prose; only then select the target expression.

5. primary object. The protected or designated object used as primary in a particular analysis. In translation, identify where the term appears and whether it is a data field, an analytic concept, a policy label or explanatory prose; only then select the target expression.

6. secondary object. The other object in that analyzed conjunction pair. In translation, identify where the term appears and whether it is a data field, an analytic concept, a policy label or explanatory prose; only then select the target expression.

7. catalog identifier. Identifier assigned within a tracking or cataloguing system. In translation, identify where the term appears and whether it is a data field, an analytic concept, a policy label or explanatory prose; only then select the target expression.

8. international designator. Launch-associated identification convention, not necessarily the same as a catalog ID. In translation, identify where the term appears and whether it is a data field, an analytic concept, a policy label or explanatory prose; only then select the target expression.

9. ephemeris. Time-referenced information describing an object’s orbit or position over time. In translation, identify where the term appears and whether it is a data field, an analytic concept, a policy label or explanatory prose; only then select the target expression.

10. orbit determination. Estimation of orbital state from observations using models. In translation, identify where the term appears and whether it is a data field, an analytic concept, a policy label or explanatory prose; only then select the target expression.

11. propagation. Prediction of orbital state at another time using a model. In translation, identify where the term appears and whether it is a data field, an analytic concept, a policy label or explanatory prose; only then select the target expression.

12. screening. Search process for possible close approaches under specified criteria. In translation, identify where the term appears and whether it is a data field, an analytic concept, a policy label or explanatory prose; only then select the target expression.

13. risk assessment. Evaluation of the encounter’s possible significance from available evidence. In translation, identify where the term appears and whether it is a data field, an analytic concept, a policy label or explanatory prose; only then select the target expression.

14. mitigation. Action or plan intended to reduce a specified risk, requiring competent authority. In translation, identify where the term appears and whether it is a data field, an analytic concept, a policy label or explanatory prose; only then select the target expression.

15. covariance. Statistical representation of variances and correlations among state components. In translation, identify where the term appears and whether it is a data field, an analytic concept, a policy label or explanatory prose; only then select the target expression.

16. variance. Squared statistical dispersion measure; distinct from a standard deviation. In translation, identify where the term appears and whether it is a data field, an analytic concept, a policy label or explanatory prose; only then select the target expression.

17. correlation. Relationship between variations of parameters under a statistical model. In translation, identify where the term appears and whether it is a data field, an analytic concept, a policy label or explanatory prose; only then select the target expression.

18. RTN. Radial–Transverse–Normal local orbital frame terminology. In translation, identify where the term appears and whether it is a data field, an analytic concept, a policy label or explanatory prose; only then select the target expression.

19. reference frame. Defined coordinate system and conventions for interpreting vectors. In translation, identify where the term appears and whether it is a data field, an analytic concept, a policy label or explanatory prose; only then select the target expression.

20. state vector. Set of position and velocity components at a specified epoch. In translation, identify where the term appears and whether it is a data field, an analytic concept, a policy label or explanatory prose; only then select the target expression.

21. epoch. Reference time associated with an orbit state or element set. In translation, identify where the term appears and whether it is a data field, an analytic concept, a policy label or explanatory prose; only then select the target expression.

22. time scale. Convention for representing time, such as UTC in relevant files. In translation, identify where the term appears and whether it is a data field, an analytic concept, a policy label or explanatory prose; only then select the target expression.

23. UTC. Coordinated Universal Time; explicitly label any local display conversion. In translation, identify where the term appears and whether it is a data field, an analytic concept, a policy label or explanatory prose; only then select the target expression.

24. creation date. When a message was generated, not automatically when an encounter happens. In translation, identify where the term appears and whether it is a data field, an analytic concept, a policy label or explanatory prose; only then select the target expression.

25. screening window. Time interval searched for possible conjunctions. In translation, identify where the term appears and whether it is a data field, an analytic concept, a policy label or explanatory prose; only then select the target expression.

26. prediction horizon. Future interval over which an estimate is considered. In translation, identify where the term appears and whether it is a data field, an analytic concept, a policy label or explanatory prose; only then select the target expression.

27. probability of collision. Model-based probability measure for a defined encounter under given assumptions. In translation, identify where the term appears and whether it is a data field, an analytic concept, a policy label or explanatory prose; only then select the target expression.

28. Pc. Common abbreviation for collision probability, dependent on defined methodology. In translation, identify where the term appears and whether it is a data field, an analytic concept, a policy label or explanatory prose; only then select the target expression.

29. HBR. Hard-body radius used in specified collision geometry assumptions. In translation, identify where the term appears and whether it is a data field, an analytic concept, a policy label or explanatory prose; only then select the target expression.

30. relative velocity. Velocity of one object relative to another. In translation, identify where the term appears and whether it is a data field, an analytic concept, a policy label or explanatory prose; only then select the target expression.

31. relative position. Position difference evaluated within a declared reference frame. In translation, identify where the term appears and whether it is a data field, an analytic concept, a policy label or explanatory prose; only then select the target expression.

32. scientific notation. Compact exponential representation such as 2E-5. In translation, identify where the term appears and whether it is a data field, an analytic concept, a policy label or explanatory prose; only then select the target expression.

33. unit. Measurement dimension indicator, including metres, kilometres or seconds. In translation, identify where the term appears and whether it is a data field, an analytic concept, a policy label or explanatory prose; only then select the target expression.

34. precision. Numerical or temporal detail carried by a representation. In translation, identify where the term appears and whether it is a data field, an analytic concept, a policy label or explanatory prose; only then select the target expression.

35. revision. Updated version of a source record or analysis. In translation, identify where the term appears and whether it is a data field, an analytic concept, a policy label or explanatory prose; only then select the target expression.

36. superseded. Status of a record replaced by a newer authoritative version. In translation, identify where the term appears and whether it is a data field, an analytic concept, a policy label or explanatory prose; only then select the target expression.

37. retracted. Withdrawn or corrected status that must remain visible in history. In translation, identify where the term appears and whether it is a data field, an analytic concept, a policy label or explanatory prose; only then select the target expression.

38. issuer. Entity that originated or formally released the message. In translation, identify where the term appears and whether it is a data field, an analytic concept, a policy label or explanatory prose; only then select the target expression.

39. operator. Entity responsible for spacecraft operations under its authority. In translation, identify where the term appears and whether it is a data field, an analytic concept, a policy label or explanatory prose; only then select the target expression.

40. recommendation. Suggested action rather than a proved completed or approved act. In translation, identify where the term appears and whether it is a data field, an analytic concept, a policy label or explanatory prose; only then select the target expression.

41. approval. Formal decision under a relevant process, not equivalent to analysis. In translation, identify where the term appears and whether it is a data field, an analytic concept, a policy label or explanatory prose; only then select the target expression.

42. commanded. A command has been issued, not necessarily executed or verified. In translation, identify where the term appears and whether it is a data field, an analytic concept, a policy label or explanatory prose; only then select the target expression.

43. executed. An action is reported as carried out, requiring actual evidence. In translation, identify where the term appears and whether it is a data field, an analytic concept, a policy label or explanatory prose; only then select the target expression.

44. verified. A state has been checked under a stated method or authority. In translation, identify where the term appears and whether it is a data field, an analytic concept, a policy label or explanatory prose; only then select the target expression.

45. TLE. Traditional two-line orbital element representation with fixed-width limitations. In translation, identify where the term appears and whether it is a data field, an analytic concept, a policy label or explanatory prose; only then select the target expression.

46. OMM. Orbit Mean-Elements Message representation in the CCSDS orbit-data family. In translation, identify where the term appears and whether it is a data field, an analytic concept, a policy label or explanatory prose; only then select the target expression.

47. GP data. General perturbations orbital data used with the appropriate model context. In translation, identify where the term appears and whether it is a data field, an analytic concept, a policy label or explanatory prose; only then select the target expression.

48. SGP4. Simplified General Perturbations 4 propagator context associated with GP elements. In translation, identify where the term appears and whether it is a data field, an analytic concept, a policy label or explanatory prose; only then select the target expression.

49. KVN. Key-value notation for particular structured data representations. In translation, identify where the term appears and whether it is a data field, an analytic concept, a policy label or explanatory prose; only then select the target expression.

50. XML. Extensible Markup Language, with schema-specific element and attribute semantics. In translation, identify where the term appears and whether it is a data field, an analytic concept, a policy label or explanatory prose; only then select the target expression.

51. JSON. Structured data representation whose keys and types must remain well defined. In translation, identify where the term appears and whether it is a data field, an analytic concept, a policy label or explanatory prose; only then select the target expression.

52. CSV. Comma-separated values; delimiter and numeric locale need explicit control. In translation, identify where the term appears and whether it is a data field, an analytic concept, a policy label or explanatory prose; only then select the target expression.

53. parser. Software that reads structured data according to a defined grammar. In translation, identify where the term appears and whether it is a data field, an analytic concept, a policy label or explanatory prose; only then select the target expression.

54. schema. Rules defining allowed data fields, types and relationships. In translation, identify where the term appears and whether it is a data field, an analytic concept, a policy label or explanatory prose; only then select the target expression.

55. null. Explicit absence marker in some data formats, not the same as numeric zero. In translation, identify where the term appears and whether it is a data field, an analytic concept, a policy label or explanatory prose; only then select the target expression.

56. missing. Data not supplied or unavailable in the current record. In translation, identify where the term appears and whether it is a data field, an analytic concept, a policy label or explanatory prose; only then select the target expression.

57. zero. A reported numeric value that must not be confused with missing. In translation, identify where the term appears and whether it is a data field, an analytic concept, a policy label or explanatory prose; only then select the target expression.

58. data provenance. The origin and transformation history of a piece of evidence. In translation, identify where the term appears and whether it is a data field, an analytic concept, a policy label or explanatory prose; only then select the target expression.

59. data vintage. Age or reference time of the data used in an estimate. In translation, identify where the term appears and whether it is a data field, an analytic concept, a policy label or explanatory prose; only then select the target expression.

60. source snapshot. Saved version of a source artifact against which translation was performed. In translation, identify where the term appears and whether it is a data field, an analytic concept, a policy label or explanatory prose; only then select the target expression.

61. translation memory. Reuse system for previously translated segments requiring variable and status checks. In translation, identify where the term appears and whether it is a data field, an analytic concept, a policy label or explanatory prose; only then select the target expression.

62. termbase. Managed terminology system with definitions, context and preferred expressions. In translation, identify where the term appears and whether it is a data field, an analytic concept, a policy label or explanatory prose; only then select the target expression.

63. quality gate. Check with a defined pass, hold or correction outcome. In translation, identify where the term appears and whether it is a data field, an analytic concept, a policy label or explanatory prose; only then select the target expression.

64. release receipt. Evidence of what was reviewed, approved and published. In translation, identify where the term appears and whether it is a data field, an analytic concept, a policy label or explanatory prose; only then select the target expression.

65. public summary. Reader-friendly account that retains the constraints of the technical source. In translation, identify where the term appears and whether it is a data field, an analytic concept, a policy label or explanatory prose; only then select the target expression.

66. incident chronology. Time-ordered record of observations, interpretations, actions and outcomes. In translation, identify where the term appears and whether it is a data field, an analytic concept, a policy label or explanatory prose; only then select the target expression.

67. space situational awareness. Knowledge derived from tracking and analysis of space objects and environment. In translation, identify where the term appears and whether it is a data field, an analytic concept, a policy label or explanatory prose; only then select the target expression.

68. space debris. Human-made orbital material not functioning as an active spacecraft, as classified by source. In translation, identify where the term appears and whether it is a data field, an analytic concept, a policy label or explanatory prose; only then select the target expression.

69. reentry. Passage into an atmosphere, a distinct problem from on-orbit conjunction risk. In translation, identify where the term appears and whether it is a data field, an analytic concept, a policy label or explanatory prose; only then select the target expression.

70. collision avoidance. Risk-management activities intended to prevent physical collision. In translation, identify where the term appears and whether it is a data field, an analytic concept, a policy label or explanatory prose; only then select the target expression.

71. uncertainty. Limit on what the data and model can establish, not permission to invent a value. In translation, identify where the term appears and whether it is a data field, an analytic concept, a policy label or explanatory prose; only then select the target expression.

72. model assumption. Parameter or condition on which a calculation or interpretation depends. In translation, identify where the term appears and whether it is a data field, an analytic concept, a policy label or explanatory prose; only then select the target expression.

73. source-to-target invariant. Fact or relationship that must remain unchanged across language transfer. In translation, identify where the term appears and whether it is a data field, an analytic concept, a policy label or explanatory prose; only then select the target expression.

Frequently asked questions on satellite conjunction translation

Question 1: Does a conjunction mean two satellites collide?

No. In this context it identifies a predicted close approach that may warrant further assessment. It is not itself a statement of physical contact. For a real document, the short answer is a starting point only: check the relevant source standard and edition, the issuing provider’s methodology and the record’s latest status. Do not fill gaps with generic assumptions simply because a sentence needs to sound complete.

Question 2: What is a CDM in simple language?

A Conjunction Data Message is a structured way of conveying information relevant to a predicted close approach, including identities, time and analytical quantities. For a real document, the short answer is a starting point only: check the relevant source standard and edition, the issuing provider’s methodology and the record’s latest status. Do not fill gaps with generic assumptions simply because a sentence needs to sound complete.

Question 3: Can a translator calculate collision probability?

Not from language expertise alone. The correct job is to preserve and explain the source estimate and route analytical questions to qualified professionals. For a real document, the short answer is a starting point only: check the relevant source standard and edition, the issuing provider’s methodology and the record’s latest status. Do not fill gaps with generic assumptions simply because a sentence needs to sound complete.

Question 4: Why does covariance matter?

It conveys the uncertainty and relationships among estimated state variables. It cannot generally be replaced by one invented single-number error margin. For a real document, the short answer is a starting point only: check the relevant source standard and edition, the issuing provider’s methodology and the record’s latest status. Do not fill gaps with generic assumptions simply because a sentence needs to sound complete.

Question 5: What does TCA mean?

Time of closest approach in the modeled encounter. It is not automatically a confirmed collision time or a recommended maneuver time. For a real document, the short answer is a starting point only: check the relevant source standard and edition, the issuing provider’s methodology and the record’s latest status. Do not fill gaps with generic assumptions simply because a sentence needs to sound complete.

Question 6: Why must catalog identifiers remain unchanged?

They are data joins for specific objects. Translated nicknames or clipped numbers can silently point to a different record. For a real document, the short answer is a starting point only: check the relevant source standard and edition, the issuing provider’s methodology and the record’s latest status. Do not fill gaps with generic assumptions simply because a sentence needs to sound complete.

Question 7: What changed with satellite catalog numbers in 2026?

CelesTrak says traditional five-digit TLE coverage does not include newly cataloged six-digit IDs after the July 2026 boundary. Use a supported alternative delivery such as OMM-compatible formats where applicable. For a real document, the short answer is a starting point only: check the relevant source standard and edition, the issuing provider’s methodology and the record’s latest status. Do not fill gaps with generic assumptions simply because a sentence needs to sound complete.

Question 8: Is the traditional TLE format completely gone?

No. Legacy TLE representations remain useful for eligible older identifiers, but they are not a universal representation for new six-digit objects. For a real document, the short answer is a starting point only: check the relevant source standard and edition, the issuing provider’s methodology and the record’s latest status. Do not fill gaps with generic assumptions simply because a sentence needs to sound complete.

Question 9: Does OMM mean the orbit is automatically accurate?

No. A modern format supports representation and metadata; accuracy depends on underlying observations, models, epoch and processing. For a real document, the short answer is a starting point only: check the relevant source standard and edition, the issuing provider’s methodology and the record’s latest status. Do not fill gaps with generic assumptions simply because a sentence needs to sound complete.

Question 10: Can I translate machine field names?

Usually not inside a standardized payload. Translate adjacent explanations and user-facing labels, but preserve protocol tokens. For a real document, the short answer is a starting point only: check the relevant source standard and edition, the issuing provider’s methodology and the record’s latest status. Do not fill gaps with generic assumptions simply because a sentence needs to sound complete.

Question 11: Should every probability be converted to percent?

Not necessarily. If a conversion helps readers, preserve the raw value, verify the calculation and label the new presentation. For a real document, the short answer is a starting point only: check the relevant source standard and edition, the issuing provider’s methodology and the record’s latest status. Do not fill gaps with generic assumptions simply because a sentence needs to sound complete.

Question 12: Can a public summary say a satellite is safe?

It should report the assessment and its scope without inventing a universal guarantee from a model-dependent estimate. For a real document, the short answer is a starting point only: check the relevant source standard and edition, the issuing provider’s methodology and the record’s latest status. Do not fill gaps with generic assumptions simply because a sentence needs to sound complete.

Question 13: Can I combine two providers’ assessments?

Compare them in separate sourced rows. Do not merge values into a synthetic event record unless a qualified, authorized process has done so. For a real document, the short answer is a starting point only: check the relevant source standard and edition, the issuing provider’s methodology and the record’s latest status. Do not fill gaps with generic assumptions simply because a sentence needs to sound complete.

Question 14: Why are creation time and TCA different?

A message is generated at a particular time about an encounter predicted at another time. The two roles are distinct. For a real document, the short answer is a starting point only: check the relevant source standard and edition, the issuing provider’s methodology and the record’s latest status. Do not fill gaps with generic assumptions simply because a sentence needs to sound complete.

Question 15: What if the source omits probability?

Write that it is not reported; do not replace absence with zero or with an inferred calculation. For a real document, the short answer is a starting point only: check the relevant source standard and edition, the issuing provider’s methodology and the record’s latest status. Do not fill gaps with generic assumptions simply because a sentence needs to sound complete.

Question 16: Does a maneuver recommendation authorize a burn?

No. Recommendations, approvals, commands, execution and post-action verification are separate states controlled by authorized mission processes. For a real document, the short answer is a starting point only: check the relevant source standard and edition, the issuing provider’s methodology and the record’s latest status. Do not fill gaps with generic assumptions simply because a sentence needs to sound complete.

Question 17: Are the numbers in this article actual alerts?

No. All teaching cases are fictional and intentionally non-operational. For a real document, the short answer is a starting point only: check the relevant source standard and edition, the issuing provider’s methodology and the record’s latest status. Do not fill gaps with generic assumptions simply because a sentence needs to sound complete.

Question 18: Which source explains risk stages?

NASA CARA describes conjunction prediction, close-approach risk assessment and risk mitigation as distinct activities. For a real document, the short answer is a starting point only: check the relevant source standard and edition, the issuing provider’s methodology and the record’s latest status. Do not fill gaps with generic assumptions simply because a sentence needs to sound complete.

Question 19: How should a termbase be structured?

Include concept, definition, source, edition, language, approved equivalent, forbidden misleading equivalents and validation examples. For a real document, the short answer is a starting point only: check the relevant source standard and edition, the issuing provider’s methodology and the record’s latest status. Do not fill gaps with generic assumptions simply because a sentence needs to sound complete.

Question 20: How do I check a translation produced by AI?

Check every identifier, time, unit, sign, exponent, qualifier, missing value, status and authority against source records, then obtain domain and language review. For a real document, the short answer is a starting point only: check the relevant source standard and edition, the issuing provider’s methodology and the record’s latest status. Do not fill gaps with generic assumptions simply because a sentence needs to sound complete.

Question 21: What if a source warning is updated?

Publish a dated correction or replacement and preserve the revision link rather than continuing to present stale text as current. For a real document, the short answer is a starting point only: check the relevant source standard and edition, the issuing provider’s methodology and the record’s latest status. Do not fill gaps with generic assumptions simply because a sentence needs to sound complete.

Question 22: Can a diagram replace the structured record?

No. Diagrams can communicate geometry or trends, but the machine-readable source remains necessary for exact values and assumptions. For a real document, the short answer is a starting point only: check the relevant source standard and edition, the issuing provider’s methodology and the record’s latest status. Do not fill gaps with generic assumptions simply because a sentence needs to sound complete.

Question 23: What does ‘not to scale’ protect?

It prevents readers treating an illustrative separation or size as a measured value. For a real document, the short answer is a starting point only: check the relevant source standard and edition, the issuing provider’s methodology and the record’s latest status. Do not fill gaps with generic assumptions simply because a sentence needs to sound complete.

Question 24: How should right-to-left languages handle identifiers?

Keep machine tokens intact and test both their rendering and copy/paste behavior so punctuation or digit order is not altered. For a real document, the short answer is a starting point only: check the relevant source standard and edition, the issuing provider’s methodology and the record’s latest status. Do not fill gaps with generic assumptions simply because a sentence needs to sound complete.

Question 25: Where does this article sit in the wider architecture?

It is the specialist node for conjunction, orbital data and multilingual evidence integrity. The canonical root owns general translation principles, and the existing space-operations child owns the broader spacecraft field. For a real document, the short answer is a starting point only: check the relevant source standard and edition, the issuing provider’s methodology and the record’s latest status. Do not fill gaps with generic assumptions simply because a sentence needs to sound complete.

Question 26: What is the best short translation QA test?

Ask whether a competent reader would infer the same object, event, timestamp, numerical quantity, degree of certainty and decision authority from both versions. For a real document, the short answer is a starting point only: check the relevant source standard and edition, the issuing provider’s methodology and the record’s latest status. Do not fill gaps with generic assumptions simply because a sentence needs to sound complete.

Research notes, live-source cautions and standards

This guide uses authoritative sources for terminology and for the distinction between orbital observation, prediction, assessment and action. Source status matters. On 31 October 2026, NASA’s CARA material describes prediction, assessment and mitigation as separate processes; its handbook page includes an appendix update in August 2026. CelesTrak documents a 9 May 2026 CSV default for GP query output and a July 11, 2026 exhaustion of the legacy five-digit catalog-number representation for newly catalogued objects. ESA’s 14 September 2026 environment report covers data through the end of 2025; it should not be presented as measurements through the later publication date. Treat new document editions and mission-specific rules as live variables requiring source verification.

1. NASA Conjunction Assessment and CARA. NASA explains three stages: conjunction prediction, risk assessment and risk mitigation, making clear that screening is not itself a collision decision.

2. NASA CARA overview. NASA describes governance, operations and risk-reduction responsibilities in spacecraft conjunction assessment.

3. CelesTrak general perturbations formats. CelesTrak documents OMM-compatible KVN, XML, JSON and CSV alongside legacy TLE formats, including 2026 identifier and default-format changes.

4. CelesTrak current GP catalogue notice. The published notice reports exhaustion of five-digit catalog identifiers on 11 July 2026 and the resulting coverage limitation of traditional TLE output.

5. ESA Space Environment Report 2026. Published 14 September 2026, using observations through 2025, this report gives context for the increasing density and complexity of the orbital environment.

6. CCSDS publication programme. Consult the official publication catalogue for the applicable issue of Conjunction Data Message and Orbit Data Messages; a draft or pink book is not automatically an adopted blue-book version.

7. ISO 5060:2024 translation evaluation. ISO provides guidance for analytical human evaluation of translated output, including error typology and sampling.

8. ISO 17100:2015 translation services. ISO describes core translation service process and resource requirements; it does not approve mission operations or replace domain review.

A note on CCSDS editions: consult the official CCSDS publication catalogue and the actual issuer’s stated data-format version. Draft proposals and final adopted standards have different normative status. This article intentionally does not assume that a proposed CDM edition applies to every producer, nor does it quote proprietary text from standards. Likewise, ISO translation guidance informs evaluation methods; it is not a source of orbital mechanics or permission to perform a maneuver.

What comes next: routes without replacing protected owners

Return to the Master Art of Translation | The Complete System for Moving Meaning Between Languages for the parent model of translation units, equivalence, context, terminology, verification and meaning preservation. This child stays specifically with orbital conjunction records and the language–evidence boundary. For the wider mission-system domain—spacecraft telemetry, command statuses, orbit operations and remote-sensing outputs—continue through Master Art of Translation | The Space Missions & Satellite Operations Translation System. The general space-operations owner remains authoritative for that broader scope.

For building precise lexical knowledge across contexts, the Vocabulary Learning Hub remains the protected owner. It is relevant when a reader needs to learn concept families, register, definitions and durable retrieval of terms. For English clauses, modality, negation, attribution, references and reading comprehension, use How English Works; the English owner remains separate. This article neither re-creates their programmes nor substitutes a satellite glossary for language learning.

One final practical test: can a reader recover which pair was analyzed, when the message and event occurred, which values were observed or predicted, how uncertainty was represented, what changed between versions, and which human authority—if any—approved an action? If the answer is yes in both source and target languages, the translation has preserved the evidence rather than merely its appearance. If not, correct the specific lost relationship, record the repair and release only after the appropriate reviewer can verify the same meaning.