MathematicsOS Glossary and Reading Map
The Controlled Language for Mathematics, Additional Mathematics, Learning Science and Civilisation
Article ID: MATHCIV-009
Canonical owner: eduKateSG
Family: Foundation
Profile: Hub
Update class: Evergreen
Quick Read
MathematicsOS uses a controlled vocabulary because several ideas that sound similar are not the same thing.
A mark is not the whole of mathematical capability.
A correct answer is not necessarily learning.
Content delivered is not necessarily content received.
Something understood today is not necessarily retained next month.
Performance on a familiar question is not proof of transfer.
AI-assisted success is not automatically independent capability.
A city average is not the experience of every person living in the city.
A mathematical model is not reality.
And an architectural analogy is not automatically a scientific law.
The central chain is:
Reality → Sense → State → Capability → Constraint → Strategy → Control → Learning → Meta-Control
For education, another important chain is:
Delivery → Receipt → Retention → Transfer → Independence → Execution
And MathematicsOS itself operates approximately as:
Sense → Index → Map → Locate → Protect → Select → Teach → Practise → Verify → Writeback → Meta-Control → Regenerate
This glossary explains what these terms mean, what they do not mean, and where to read next.
1. Start Here: Which Route Do You Need?
The Mathematics and Civilisation estate contains 100 planned articles. You do not need to read them sequentially.
Start from the problem you are trying to understand.
Route A — “I want to understand a learner”
Begin with:
- MATHCIV-005 — From Marks to Mathematical Capability
- MATHCIV-007 — The Mathematics Capability Vector
- MATHCIV-025 — MathematicsOS: A Complete Learner Capability Runtime
- MATHCIV-026 — The Mathematics Prerequisite Graph
- MATHCIV-027 — The Moving Bottleneck Set
- MATHCIV-033 — The Mathematics Error Ledger
- MATHCIV-035 — The Independence Gate
- MATHCIV-036 — A Mathematics Learner State Vector and Dashboard
This route is useful when a student appears to be “weak at maths”, but the actual mechanism is unclear.
The point is to replace a vague identity label with a better question:
What is currently stable, what is fragile, what is missing, what is support-dependent, and what evidence would distinguish the competing explanations?
Route B — “I want the learning science”
Read MATHCIV-010 to MATHCIV-024.
This cluster covers:
working memory, cognitive load, worked examples, retrieval, spacing, interleaving, productive failure, error analysis, metacognition, anxiety, emotion, representations, real-world transfer and neuroscience.
Start with the mechanism you are interested in rather than looking for one universally “best” teaching method.
MathematicsOS assumes that instructional effectiveness depends on the Receiver, starting State, target capability, task and conditions.
Route C — “I need Additional Mathematics”
Start with:
- MATHCIV-040 — Additional Mathematics Tuition in Singapore: The Capability-Building Guide
- MATHCIV-045 — Is a Student Ready for Additional Mathematics?
- MATHCIV-047 — Secondary 3 Additional Mathematics
- MATHCIV-048 — Secondary 4 Additional Mathematics
- MATHCIV-049 — Algebra for Additional Mathematics: The Integrity Gate
- MATHCIV-057 — Mixed Additional Mathematics Questions
- MATHCIV-058 — Additional Mathematics Exam Preparation
Then move into the topic articles from MATHCIV-050 onward.
Current examination terminology belongs in MATHCIV-041 to MATHCIV-043 and must be reverified against official sources rather than copied indefinitely from older pages.
Route D — “I want mathematics and civilisation”
Read MATHCIV-060 to MATHCIV-084.
These articles move beyond classroom mathematics into:
measurement, standards, accounting, maps, geometry, algebra, probability, calculus, statistics, logistics, algorithms, economic complexity, adult numeracy, cities, networks, infrastructure and mathematical governance.
The governing boundary is important:
Mathematics can enable civilisation capability without being a sufficient explanation for civilisation.
Route E — “I want AI, mathematical reasoning and verification”
Read MATHCIV-085 to MATHCIV-092.
This cluster covers AI tutors, guardrails, human–AI tutoring, knowledge tracing, automated theorem proving, mathematical trust, AI dependence and what curricula may need to protect.
For research standards, evidence control and system maintenance, continue into MATHCIV-093 to MATHCIV-099.
2. Reality and Time
Reality
Reality is what exists or occurs independently of the model being used to describe it.
A test score is not Reality.
A dashboard is not Reality.
A tutor’s judgment is not Reality.
A knowledge-tracing probability is not Reality.
They may all contain useful information about Reality, but they remain observations, estimates or models.
This distinction protects MathematicsOS from an important error:
The map must never silently become the territory.
Time
Time matters because capability changes.
A student can know something today and forget it later.
A city can appear stable while underlying dependencies deteriorate.
An intervention can work immediately but fail under delayed testing.
For this reason, MathematicsOS attaches observations to a time horizon.
“Can solve quadratics” is incomplete.
A better statement is:
“Can independently solve this family of quadratic problems, under these conditions, at this time, with this level of delayed retention and transfer.”
3. The Adaptive Unit Gate
An Adaptive Unit is a system that meaningfully senses, updates its state, acts, receives feedback, maintains memory, learns and can modify aspects of its own control policy.
Not everything qualifies.
Mathematics itself is a knowledge and representation system.
An equation is a symbolic object.
Neither is automatically an adaptive unit.
A learner may qualify.
A tutor may qualify.
A school may qualify as an adaptive institution if feedback, memory, decision and organisational learning actually operate.
A classroom is a coupled multi-agent system, but the students do not merge into one mind.
A city is a multi-agent field, not one giant individual.
A civilisation is a multi-scale field of people, institutions, technologies, environments and inherited structures.
This distinction becomes important whenever CivilisationOS language is used.
4. Purpose
Purpose describes what the adaptive system is attempting to preserve, create or accomplish.
In MathematicsOS the purpose is not simply:
“Produce more correct answers.”
The stronger target is:
Build accurate, transferable, independently verifiable mathematical capability while protecting wellbeing, time, prerequisite integrity and future learning capacity.
Purpose must therefore remain separate from a convenient proxy.
A grade can matter.
It should not silently replace the entire purpose.
5. Receiver
The Receiver is where the useful effect must actually become real.
In education, the primary Receiver is usually the learner.
A school can deliver a lesson.
A tutor can explain a method.
An AI system can generate a beautiful solution.
None of these proves that the Receiver has acquired the capability.
This creates one of the central MathematicsOS rules:
Optimise receiver-realised capability, not merely producer effort or delivered volume.
The Receiver principle also scales beyond education.
A city policy can look efficient at system level while exporting cost, time, stress or risk to particular residents.
The average improvement is therefore not the whole story.
6. Sense
Sense means gathering information about the present system.
For a mathematics learner, useful sensing can include:
- actual work;
- first moves;
- full working;
- errors;
- latency;
- confidence;
- hint use;
- representation choice;
- transfer performance;
- retention;
- affect or avoidance signals;
- timing;
- support state.
Sensing should collect evidence before producing labels.
“Careless student” is not a sensor reading.
“Lost the negative sign during the third transformation and did not detect the inconsistency during checking” is an observation.
The second is far more useful.
7. State
State is the current measured configuration of the system.
It answers:
What appears to be true now?
State is not Capability.
A learner can currently answer a familiar question correctly while lacking the ability to transfer the method to a new representation.
Conversely, a learner may possess useful mathematical capability but temporarily perform poorly because of unfamiliar notation, time pressure, misunderstanding of the question or another constraint.
Hence:
State ≠ Capability
This distinction is foundational to MathematicsOS.
8. Capability
Capability is the demonstrated ability to produce a required mathematical outcome under declared conditions.
It is stronger than one successful performance.
Stable capability generally requires evidence across:
- correctness;
- retention;
- transfer;
- independence;
- verification;
- realistic execution conditions.
Capability must therefore be indexed.
Not:
“The student is good at algebra.”
But:
“The student currently demonstrates stable independent control of these algebraic transformations across these task families and representations.”
9. The Mathematics Capability Vector
MathematicsOS replaces the single scalar idea of being “good at maths” with a multidimensional capability vector.
CON — Conceptual Structure
Understanding the relationships that make the mathematics work.
Not merely remembering a sequence of moves.
PRO — Procedural Accuracy
Executing mathematical procedures correctly and reliably.
Conceptual understanding and procedural skill should not be treated as permanent rivals. They can strengthen one another.
SYM — Symbolic Control
Maintaining valid transformations across algebraic expressions and equations.
This includes signs, restrictions, equivalence, variables and conditions.
REP — Representation
Constructing and translating between:
- words;
- equations;
- tables;
- graphs;
- diagrams;
- geometric representations;
- contextual models.
Knowing each representation separately is not enough if the learner cannot move between them.
SEL — Strategy or Route Selection
Choosing an appropriate method.
A student may know several mathematical procedures yet still fail mixed questions because the difficult operation is deciding which method applies now.
TRN — Transfer
Using a mathematical idea when the surface appearance changes.
Transfer may involve a change in:
- wording;
- representation;
- context;
- parameters;
- direction of reasoning;
- combination with another topic.
Transfer should be tested rather than assumed.
MOD — Modelling
Turning a real or conceptual situation into mathematical structure, while keeping assumptions and interpretation visible.
A model can be mathematically correct and still poorly represent Reality.
VER — Verification
Checking whether an answer, transformation, model or conclusion survives scrutiny.
Verification may involve:
- substitution;
- alternative methods;
- inverse operations;
- estimation;
- units;
- signs;
- domains;
- graphical behaviour;
- formal proof.
Verification becomes particularly important when AI can generate plausible-looking mathematics at scale.
FLU — Fluency
Accurate, increasingly efficient mathematical execution.
Fluency is not raw speed.
The relevant standard is closer to:
accuracy + route quality + readable working + recovery + appropriate speed
RET — Retention
Ability to reconstruct the capability after time has passed.
Immediate post-teaching performance is weak evidence of long-term retention.
IND — Independence
Ability to perform without the tutor, answer key, peer, scaffold or generative AI supplying the target operation.
Independence is one of the most important capability gates in an AI-rich learning environment.
REG — Regulation
The metacognitive, motivational and emotional regulation required to use mathematical knowledge effectively.
This includes monitoring, checking, persistence, calibration and recovery.
REG should not be confused with diagnosing personality, intelligence or mental-health conditions.
10. Constraint
A Constraint limits what the Receiver can currently do.
Constraints may include:
- missing prerequisites;
- unstable procedures;
- representation difficulty;
- time;
- cognitive load;
- anxiety;
- poor route selection;
- weak checking;
- support dependence;
- instructional mismatch.
The important point is that a constraint should be evidenced rather than guessed.
11. Strategy
Strategy is the selected approach for moving from the current State toward the desired capability.
Different receivers may need different strategies even when the visible incorrect answer looks identical.
One learner may need a prerequisite repair.
Another may need representation translation.
Another may understand everything except route selection.
Another may simply require delayed retrieval.
This is why MathematicsOS resists one-method-for-everyone solutions.
12. Control
Control is the process of applying an intervention and checking whether the system moves in the intended direction.
In education, this might mean:
diagnose → intervene → observe → verify → update.
Control is not domination.
Nor does mathematical controllability prove practical feasibility.
A system may be theoretically controllable while requiring impossible information, precision, energy, authority or cost.
13. Memory
Memory is stored or preserved information.
For a learner, remembering a formula is memory.
For an institution, an archived procedure may be memory.
Memory does not automatically imply Learning.
A system can store information while continuing to make the same mistake.
14. Learning
Learning means that future behaviour or capability changes because of experience.
A stored answer is not sufficient.
A copied solution is not sufficient.
An AI conversation is not sufficient.
Learning becomes more convincing when the Receiver can later reconstruct and use the idea independently.
15. Meta-Control
Meta-Control changes the way control itself operates.
Suppose a tutor repeatedly teaches a topic using the same representation and the learner continues to fail.
Ordinary Control may say:
“Practise more.”
Meta-Control asks:
“Is our diagnosis wrong? Is the representation wrong? Is the sequence wrong? Should the support policy change?”
Meta-Control is therefore the mechanism that allows MathematicsOS to correct its own intervention strategy.
16. Coordination, Arbitration and Allocation
These are conditional multi-agent functions.
Coordination
Aligning actions among multiple agents.
For example, student, school, tutor and family may all influence the same learning schedule.
Arbitration
Resolving incompatible claims or objectives.
For example:
more practice may improve preparation but reduce sleep, recovery or time available for another subject.
Allocation Under Scarcity
Deciding how limited resources such as time, attention, tutor capacity or funding should be distributed.
These functions become important only when multiple receivers or purposes create competing claims.
17. BaseFloor
The BaseFloor is the capability that must remain protected while the system advances.
The principle is:
Do not gain frontier performance by destroying the stable base required for future independent operation.
Examples include:
- using a calculator before number sense is established;
- using AI to perform algebra the learner is supposed to learn;
- practising only exotic problems while routine prerequisites decay;
- accelerating content so aggressively that retention collapses.
The BaseFloor is especially important when new technology makes assisted performance very easy.
18. P3 — Stable Functioning
P3 means stable, usable operation.
For mathematics this includes:
- accurate fundamentals;
- independent execution;
- readable working;
- delayed retention;
- route selection;
- transfer across common variants;
- checking;
- recovery;
- realistic execution.
P3 is not “basic” in the sense of low ambition.
It is the stable platform from which advanced work becomes meaningful.
19. P4 — Protected Frontier Extension
P4 is advanced exploration built on top of P3.
Examples include:
- unfamiliar problems;
- mathematical synthesis;
- proof;
- richer modelling;
- advanced representations;
- computational exploration;
- AI-assisted comparison;
- research questions.
The rule is:
P4 may extend P3. It may not cannibalise the P3 operation being learned.
If AI solves the algebra, generates the plan and checks the answer, the output may be impressive while the learner’s own capability remains unknown.
20. Delivery, Receipt, Retention, Transfer, Independence and Execution
These terms form one of the most important chains in the entire MathematicsOS architecture.
Delivery
Was content presented?
Receipt
Did the Receiver attend to it and make enough sense of it to construct a usable representation?
Retention
Can it be reconstructed after a delay?
Transfer
Can it be used when something important about the task changes?
Independence
Can it be performed without the support that originally produced success?
Execution
Can the learner still operate accurately, readably and recoverably under realistic load and timing?
A lesson can therefore succeed at Delivery while failing at later gates.
That is why:
Delivered content ≠ learned capability.
21. Prerequisite Graph
A Prerequisite Graph models mathematics as connected demonstrable capabilities rather than merely a sequence of textbook chapters.
A node represents a capability.
An edge indicates that one capability materially supports another under a declared task family.
The graph is not universal.
Dependencies can vary by task, curriculum, representation and Receiver.
22. Active Bottleneck Set
The Active Bottleneck Set is the group of currently binding weaknesses that most constrain downstream performance.
This replaces an overly simple idea of one permanent “root weakness”.
The bottleneck can move.
Repair algebra and the next binding problem may become representation.
Repair representation and route selection may become visible.
Therefore:
Repair → resense → remap.
23. First Invalid Step
The First Invalid Step is the earliest point in the learner’s observable working where mathematical validity breaks.
It is often more informative than the final wrong answer.
A final answer may contain five later consequences of one early error.
Repairing the final line does not necessarily repair the mechanism.
24. Error Ledger
The Error Ledger is a structured record of meaningful mathematical errors.
It records information such as:
- task;
- representation;
- support state;
- first invalid step;
- confidence;
- self-correction;
- hint depth;
- latency;
- candidate mechanism;
- repair;
- delayed retest;
- transfer outcome.
The likely mechanism remains a hypothesis until enough evidence supports it.
An error ledger should help a learner escape repeated errors, not become a permanent identity label.
25. Calibration
Calibration asks whether confidence matches actual performance.
High confidence with low accuracy is dangerous because the learner may stop checking.
Low confidence with high accuracy creates a different problem.
Confidence alone is therefore not enough.
Confidence ≠ Calibration.
26. Support State and Hint Depth
A result should record the level of help present when it was produced.
MathematicsOS uses an AI help ladder from:
H0 — no help
through progressively stronger assistance to:
H6 — full solution with decision annotations
The important distinction is that an H6 success and an H0 success are not equivalent evidence of independent capability.
27. AI-Off Test
An AI-off test checks whether capability remains after AI assistance is removed.
A strong verification sequence asks whether the learner can:
- explain the governing relation;
- solve a structurally similar fresh problem;
- handle a changed representation;
- verify the result independently;
- repeat qualifying performance after a delay.
If those conditions are not yet met, the safe record is:
Assisted performance observed; independent capability not yet established.
28. Writeback
Writeback means updating the system’s record of learner capability.
MathematicsOS does not write an improvement into the capability record merely because a learner produced a correct AI-assisted answer.
Writeback requires qualifying evidence.
This protects the system from recording rented performance as owned capability.
29. Regeneration
Regeneration goes beyond retaining knowledge.
It asks whether the capability can be reconstructed, adapted, explained and eventually transmitted to another Receiver.
At civilisation scale, regeneration matters because knowledge that exists only inside one expert, institution or machine can disappear when that carrier disappears.
Mathematical notation can help preserve knowledge.
But stored notation is not enough.
A future Receiver must still be capable of understanding and using it.
30. Observation
An Observation is what was directly measured or recorded.
Example:
“The student selected differentiation for this problem.”
31. Inference
An Inference is an interpretation drawn from observations.
Example:
“The student may be relying on a surface cue rather than recognising the underlying structure.”
That may be a good hypothesis.
It is still not the same thing as the observation.
32. Model
A Model is a structured representation used to describe, predict or control part of Reality.
Models can be very useful.
They can also omit variables, people, context and mechanisms.
A model result should therefore not automatically be treated as a real-world fact.
33. Metaphor
A Metaphor transfers structure from one domain to another to make reasoning easier.
Words such as:
- field;
- energy;
- bottleneck;
- network;
- control;
- ecosystem;
may be useful metaphors or mathematical analogies.
But analogy does not automatically create scientific equivalence.
34. Correlation and Causation
Correlation means variables change together in a statistical relationship.
Causation means changing one factor produces a change in another under the relevant causal conditions.
Correlation may suggest a mechanism worth investigating.
It does not establish one by itself.
This distinction becomes particularly important in educational, economic, city and neuroscience claims.
35. Heterogeneity
Heterogeneity means an intervention’s effect differs across populations, designs, tasks or conditions.
An average effect does not imply that every Receiver experienced that effect.
MathematicsOS therefore asks:
Effective for whom, at what State, for which target, under what conditions?
36. Evidence Labels
MathematicsOS uses controlled evidence labels.
OFFICIAL_CURRENT
An official authority for a current fact.
Current facts still require reverification when they may change.
META_STRONG
A relevant peer-reviewed systematic review or meta-analysis.
Strong evidence does not erase heterogeneity or scope limitations.
RCT_STRONG
A relevant randomised trial.
A strong causal estimate remains bounded by population, implementation, comparator and outcome.
OBS_ASSOCIATIONAL
An observational association.
Useful for describing patterns.
Not sufficient by itself for a causal claim.
EMERGING_PEER_REVIEWED
Recent peer-reviewed evidence with limited replication, narrower scope or an emerging evidence base.
WORKING_PAPER
Serious research that has not yet completed the full peer-reviewed publication process.
Its provisional status must remain visible.
FOUNDATIONAL
Older field-shaping work that remains important but should be interpreted alongside later evidence and limitations.
ARCHITECTURAL_SYNTHESIS
A reasoned system interpretation built from multiple evidence streams.
It is not the same as saying one experiment directly measured the architecture.
BLOCKED_CLAIM
A statement the available evidence does not justify publishing as fact.
“Blocked” is an important scientific outcome.
It means the system has successfully prevented overclaiming.
37. Disconfirmation
Disconfirmation means actively defining what evidence would make us weaken, modify or abandon a hypothesis.
A serious MathematicsOS claim should therefore have not only supporting evidence but also:
- rival explanations;
- falsifying observations;
- a test;
- a review point;
- a possible architecture update.
A system that can only confirm itself is not learning.
38. Universal Grammar and Local Coefficients
Universal Grammar means a recurring structural language may be useful across many domains.
For example:
Receiver, State, Capability, Constraint and Feedback.
Universal coefficients would mean claiming the same numerical relationships apply everywhere.
MathematicsOS rejects that leap.
The grammar may travel.
The coefficients must usually be estimated locally.
39. Dashboard
A Dashboard is a sensor interface.
It displays selected measurements.
It is not the learner.
It is not the classroom.
It is not the city.
Useful dashboards therefore expose:
- data freshness;
- uncertainty;
- missing probes;
- support state;
- distribution;
- competing hypotheses.
The governing rule is simple:
Dashboard ≠ Reality.
40. Receiver Distribution and Lower Tail
A Receiver Distribution asks how outcomes are distributed across individuals rather than reporting only an average.
The Lower Tail refers to receivers who are furthest from stable access to the relevant capability or benefit.
A rising average can coexist with a deteriorating lower tail.
CivilisationOS therefore requires distributional inspection.
41. Burden Export
Burden Export occurs when an intervention improves one visible metric by shifting cost to a less visible Receiver.
Educational examples may include:
- excessive time;
- sleep displacement;
- anxiety;
- family logistical load;
- opportunity cost;
- dependency on paid support.
A mathematically efficient solution can therefore still be poor system design if hidden burdens are ignored.
42. Mathematics as a Civilisation Capability Layer
Within this project, mathematics is treated as supporting seven major civilisation functions.
Observe and Measure
Make quantities and tolerances comparable.
Represent and Compress
Encode relationships in symbols, graphs, coordinates and models.
Infer and Explain
Reason from premises, evidence and uncertainty.
Coordinate and Standardise
Allow people to act consistently across distance and time.
Design and Control
Select actions under constraints.
Verify and Audit
Make claims inspectable.
Transmit and Regenerate
Preserve methods so future Receivers can reconstruct and use them.
This is an architectural synthesis.
It is not a claim that one experiment measured a universal quantity called “civilisation power”.
43. EnDist
EnDist is retained in CivilisationOS as a typed accounting and routing grammar:
Field → Composition → Coupling → Motion → Conversion → Persistence
Its safe use is architectural.
Unless a particular domain provides real physical units and validated equations, EnDist should not be described as a literal conserved social or cognitive energy.
This distinction protects useful system reasoning without manufacturing physics where none has been demonstrated.
44. Structural Controllability
Structural Controllability is a mathematical property describing whether a system can theoretically be driven through its state space under specified assumptions.
It does not automatically answer:
- whether intervention is affordable;
- whether sufficient precision exists;
- whether the necessary authority exists;
- whether the intervention is ethical;
- whether the required energy or information is realistic.
Hence:
Theoretical controllability ≠ practical control.
45. Early-Warning Signal
An Early-Warning Signal is a measurable pattern that may precede a system transition under particular mechanisms.
Possible signals in some domains include changing variance, autocorrelation or recovery behaviour.
MathematicsOS does not assume that one universal warning score predicts all learner, school, city or civilisation transitions.
Every sensor requires calibration.
46. Regime, Threshold and Tipping
A Regime is a relatively stable pattern of system behaviour.
A Threshold is a boundary at which behaviour may change substantially.
Tipping describes movement into another regime.
These concepts can be mathematically useful, but importing them from ecological or physical systems into human learning requires explicit calibration.
They should not become decorative scientific language.
47. Civilisation Math Value
The master proposes an architectural value expression:
Useful Capability
× Transfer
× Verification
× Independent Use
× Distributional Reach
× Regeneration
× Safety Factor
This is not a fitted universal equation.
Its purpose is to reveal a systems principle:
A near-zero term can collapse realised value.
For example, sophisticated mathematical output with almost no independent verification may create fragile capability.
High frontier performance with a collapsing lower tail violates the BaseFloor.
And large amounts of AI-assisted production with little human independent use may represent rented capability rather than regenerated capability.
48. The MathematicsOS Runtime
The full MathematicsOS operating loop is:
1. Sense
Collect evidence.
2. Index
Attach evidence to Receiver, topic, task, representation, time and support state.
3. Map
Update the prerequisite graph and capability vector.
4. Locate
Identify active bottlenecks and uncertainty.
5. Protect
Preserve the BaseFloor, independent operation and wellbeing.
6. Select
Choose the smallest sufficient intervention.
7. Teach
Build a valid representation.
8. Practise
Move from stabilisation toward retrieval, spacing, selection and transfer.
9. Verify
Test retention, transfer, independence and execution.
10. Writeback
Update capability only from qualifying evidence.
11. Meta-Control
Change the intervention policy when evidence contradicts the model.
12. Regenerate
Require explanation, adaptation and eventual independent reconstruction.
The runtime is deliberately cyclical.
If Reality disagrees with the diagnosis, the system returns and updates itself.
49. Reading Map by Learner State
“The student scores well but cannot handle unfamiliar questions.”
Read:
MATHCIV-005 → 007 → 016 → 023 → 031
Focus on Capability, Transfer, route selection and context dependence.
“The student understands during tuition but forgets later.”
Read:
MATHCIV-014 → 015 → 034
Focus on retrieval, spacing and delayed retention.
“The student knows individual topics but struggles with mixed questions.”
Read:
MATHCIV-016 → 030 → 031 → 057
Focus on interleaving, method selection and cross-topic bridges.
“The same type of mistake keeps returning.”
Read:
MATHCIV-018 → 027 → 033
Focus on the first invalid step, error classification and the moving bottleneck.
“The student is fast but makes too many errors.”
Read:
MATHCIV-007 → 032 → 058
Focus on FLU, verification, recovery and paper control.
“The student is anxious about mathematics.”
Read:
MATHCIV-020 → 021 → 025
Separate the emotional State from the actual mathematical capability and repair both where needed.
“AI makes the student look much better.”
Read:
MATHCIV-006 → 035 → 085 → 086 → 091
Ask what remains when AI is removed.
“We are unsure whether the student is ready for Additional Mathematics.”
Read:
MATHCIV-040 → 045 → 049
Do not infer readiness from one mark.
50. Reading Map by Scientific Question
How does the brain learn mathematics?
MATHCIV-010, 011, 024
How should cognitive load be managed?
MATHCIV-012, 013
Does retrieval help?
MATHCIV-014
Why use spacing?
MATHCIV-015
Why mix question types?
MATHCIV-016
Can struggle before instruction help?
MATHCIV-017
What can errors tell us?
MATHCIV-018, 033
How useful is confidence?
MATHCIV-019
What about mathematics anxiety?
MATHCIV-020
Do emotions matter?
MATHCIV-021
Should teachers use multiple representations?
MATHCIV-022, 028
Why does classroom knowledge fail to transfer?
MATHCIV-023, 031
51. Reading Map by Civilisation Question
Why does mathematics matter beyond school?
MATHCIV-002
How does notation preserve knowledge?
MATHCIV-060
Why are standards powerful?
MATHCIV-061
What does mathematics do for administration?
MATHCIV-062
How do maps and calendars coordinate societies?
MATHCIV-063
How does geometry shape built environments?
MATHCIV-064
Why is algebra a technology of generalisation?
MATHCIV-065
How does mathematics handle collective risk?
MATHCIV-066
Why does calculus matter for changing systems?
MATHCIV-067
What can statistics tell public decision-makers?
MATHCIV-068
How does optimisation work under scarcity?
MATHCIV-069
What changes when mathematics becomes executable computation?
MATHCIV-070
What is economic complexity?
MATHCIV-071
Why does adult numeracy matter?
MATHCIV-072
How can mathematical systems distribute burdens unfairly?
MATHCIV-073, 074
How should we mathematically think about cities?
MATHCIV-075 to MATHCIV-084
52. Five Sentences to Remember
If the terminology becomes complicated, return to these five sentences.
1. State is not Capability.
What happened on one task is not automatically what the Receiver can do generally.
2. Delivery is not Learning.
Learning requires later evidence at the Receiver.
3. Transfer must be tested.
Do not infer general capability from familiar performance.
4. Assistance must eventually fade.
Tutor and AI support are useful only if the learner increasingly owns the operation.
5. The model must remain corrigible.
When Reality disagrees with the dashboard, diagnosis or architecture, update the model rather than forcing Reality to fit it.
Final Perspective
MathematicsOS is not an attempt to make education sound more technical.
Its vocabulary exists to make reasoning more disciplined.
It prevents a mark from becoming an identity.
It prevents an explanation from becoming assumed learning.
It prevents a dashboard from becoming the learner.
It prevents AI assistance from becoming fake mastery.
It prevents a mathematical analogy from becoming an invented law of civilisation.
And it gives parents, students, educators and researchers a common language for asking a better question:
What capability has actually become real in the Receiver, under what conditions, how do we know, what remains uncertain, and what is the next useful move?
That is the purpose of the glossary.
And it is the reading map for everything that follows.