Comprehension OS (Standalone Canonical Page) — Almost-Code + V1.1 (Paste-Ready)

Comprehension_OS := a meaning-extraction and evidence-discipline system
that converts unseen text into:
  accurate literal understanding + stable inference,
while preventing guessing and misread errors under load.

P3(Comprehension) = student can consistently:
  - retrieve correct details
  - infer without over-inferencing
  - justify answers with evidence lines
  - stay stable after re-read
across varied passage types.

Most comprehension errors are system errors:
  misread, vocab gap, inference leap, evidence absence, time pressure.
Comprehension improves when you tag the error type and repair that leak.

Comprehension_OS {
  Input:
    - unseen passage
    - questions
  Process:
    1) Literal Extraction (facts, who/what/when/where)
    2) Relationship Mapping (cause/contrast/change/sequence)
    3) Inference (what is implied, not stated)
    4) Evidence Anchoring (line-based justification)
    5) Answer Construction (precise scope, no over-claim)
  ControlLayer:
    - anti-guess rule
    - re-read stability check
    - time boxing
  Output:
    - accurate answers with explainable reasoning
}

LIT = what the text explicitly says.

INF = what logically follows from the text without adding new ideas.

Most students fail because they treat INF as permission to guess.
Inference must be anchored in evidence + logic.

P0: cannot extract details reliably; random guessing
P1: can answer when guided; inconsistent alone
P2: can do well sometimes; collapses under traps/time
P3: stable accuracy + evidence discipline under exam load

FM1: Misread detail (skips a key word like “not”, “except”, “only”)
FM2: Vocabulary gap (doesn't know a key word)
FM3: Inference leap (over-infer beyond text)
FM4: Evidence absence (answer not supported by lines)
FM5: Wrong scope (turns “some” into “all”, or specific into general)
FM6: Time-pressure collapse (rushes, stops checking)

COMP_Sensors {
  LiteralAccuracy (%)
  InferenceStability (same answer after re-read? Y/N)
  EvidenceLineHabit (can point to line(s)? Y/N)
  WrongAnswerTags distribution
  TrapWordMissRate ("not/except/only" missed?)
}

StabilityAfterReRead:
  If the answer changes after re-read, the process is unstable (P2).

WrongAnswerTag :=
  MISREAD         // skipped/incorrectly parsed words
  VOCAB_GAP       // key word unknown
  OVER_INFER      // adds ideas not supported
  NO_EVIDENCE     // cannot justify with lines
  WRONG_SCOPE     // exaggeration or wrong quantifier
  TIME_RUSH       // rushed without check

Repair the dominant tag first.
Do not do “more practice” without a dominant tag.

If you cannot point to evidence,
mark the answer as uncertain and re-read.
No evidence = no confidence.

EVIDENCE_ROUTINE {
  1) choose answer
  2) underline evidence line(s)
  3) rewrite answer in your own words
}

SCOPE_CONTROL {
  check quantifiers:
    some / many / most / all
  check time scope:
    once / usually / always
  check certainty words:
    may / might / must
}

QType :=
  Detail | Reference | Vocabulary-in-context | Inference | Purpose/Tone | Summary-of-idea

Find the line(s) that answer the question.
Do not answer from memory alone.

Answer must match:
  exact detail + exact scope + evidence

Re-read relevant lines.
If answer changes → tag the reason.

LOOP_MISREAD {
  drill trap words:
    not / except / only / unless / despite / however
  exercise:
    rewrite the sentence in simpler words
  goal:
    stop skipping key logic words
}

LOOP_VOCAB_GAP {
  extract 5 key words from passage
  do:
    - simple definition
    - sentence usage
    - paraphrase replacement
  reattempt the question
}

LOOP_INFERENCE {
  rule:
    inference must be “supported, not invented”
  drill:
    write:
      Claim:
      Evidence line:
      Why it implies the claim:
}

LOOP_EVIDENCE {
  every answer must include:
    Evidence line number(s)
  if cannot find evidence:
    re-read and re-locate
}

LOOP_SCOPE {
  highlight quantifiers and certainty words
  rewrite answer with correct scope:
    some → some
    may → may (not must)
}

LOOP_TIME {
  time boxing:
    answer first pass
    reserve last 3–5 minutes to re-check
  re-check routine:
    trap words + scope words
}

Summary needs point extraction.
Point extraction depends on comprehension accuracy + relation mapping.

Vocab gaps cause comprehension collapse.
But vocab only counts if retrievable; measure and repair via Vocabulary OS.

Student does many papers
→ never tags wrong-answer type
→ repeats same misread / over-infer pattern
→ guesses again under time pressure
→ accuracy stagnates

Comprehension improves by:
  wrong-answer tagging → targeted repair loops → re-test,
not by volume alone.

COMPREHENSION_OS_SPEC v1.0 {
  Definition: meaning extraction + evidence discipline under load
  Types: literal + inferential (no guessing)
  Sensors: accuracy, stability-after-reread, evidence habit, tag distribution
  Tags: MISREAD|VOCAB_GAP|OVER_INFER|NO_EVIDENCE|WRONG_SCOPE|TIME_RUSH
  Repairs: tag-specific loops
  P3 Gate: stable answers + evidence lines + correct scope
}