Canonical ID: EducationOS.PrimaryMath.FailureAnalysis.Works.v1.2
Scope: Child (Z0) + Parent/Tutor (Z1) + School/Test (Z2)
ChronoFlight Lens: Entity = Structure × Phase × Time (weekly slices; Phase = corridor altitude)
Meaning of “failed”: the child could not hold a P2 corridor under test load (time, multi-step variation, word problems).

AI Ingestion Lock

Mechanism diagnosis + repair corridor. Not blame.
Frozen primitives only: P0–P3, Z0–Z6, Rate-Dominance, FenceOS, APRC (Truncation+Stitching), ChronoFlight, InterstellarCore (optional).

Classical Foundation Block (standard)

Primary Mathematics assesses whether a student can:

compute accurately (whole numbers, fractions, decimals, percentages),
apply methods to solve problems (especially word problems),
handle multi-step questions,
interpret diagrams/tables,
under time pressure.

Civilisation-Grade Definition (eduKateSG / CivOS)

Primary Math “works” when the child has a stable Math Pipeline:

Language → Representation → Method → Steps → Answer → Check

Your child failed when this pipeline collapsed under load, even if they can do easier questions at home.

ChronoFlight Placement (why failure appears suddenly)

Tests are higher-load flight segments:

mixed topics,
unfamiliar wording,
multi-step coupling,
time pressure.

Fail = drift beat repair during the flight.
Most Primary Math failure is not “low intelligence”—it is bind failure + weak repair loops.

Phase Ladder for Primary Math (P0–P3)

P0 (Below corridor / collapses under variation)

guesses; wrong operations; steps drift; many blanks; panic.
word problems collapse first.

P1 (Narrow corridor / works only when familiar)

can do routine sums; breaks when questions change skin or add a step.

P2 (Stable corridor / pass reliably)

consistent method choice + multi-step stability; completes paper in time.
word problems are translated correctly to models/equations.

P3 (Transfer corridor / InterstellarCore seed)

can explain reasoning, choose among methods, self-correct mid-solution.
handles unfamiliar variants by decomposing.

Core Law (Rate-Dominance) — why the child failed

Failure condition:

Ġ(repair + correction + retest + method training + timed stability) < Ḋ(drift + forgetting + confusion + time-pressure + careless cascade)

Most children fail because practice increases volume, but repair stays low:

no correction/retest loop,
weak word-problem representation,
careless errors not converted into checks,
time-to-collapse (TTC) too low.

The 8 Most Common Root Causes (what usually caused the fail)

1) Weak “Language → Representation” bind (word problems)

Child cannot turn text into a bar model / equation / steps.
Signature: computation ok, word problems fail badly.

2) Wrong method choice (operation selection)

Child chooses add instead of multiply, or mixes up steps.
Signature: correct working style but wrong operation early.

3) Multi-step chain breaks (cascade errors)

One early slip destroys the whole solution.
Signature: “almost there” but wrong final answer often.

4) Weak number sense (no reasonableness check)

Child can compute but cannot tell if answer is sensible.
Signature: absurd answers not noticed.

5) Fractions/decimals/percent foundations are shaky

These topics create invisible drift across many questions.
Signature: repeated errors in conversion/comparison.

6) Careless errors are treated as personality, not a skill

Child repeats the same “careless” mistakes because checking is not trained.
Signature: same slip types weekly; no improvement.

7) No correction + retest loop (mistakes never die)

Student reviews answers but doesn’t retest.
Signature: RetestGain(7d) ≈ 0.

8) TTC is low (time-to-collapse)

Child starts okay, then collapses near the end (rushing, blanks).
Signature: last section weak; incomplete paper.

FenceOS: Sensors that reveal the true failure mode (fast diagnosis)

Track weekly with small timed sets:

TTC (time-to-collapse on a timed mixed set)
RepresentationHitRate (% word problems modeled correctly)
MethodSelectionRate (% correct method chosen)
StepErrorRate (errors per multi-step solution)
CarelessErrorRate (avoidable slips)
CompletionRate (% paper/section finished)
RetestGain_7d
NIT (avoidance/shutdown: “I can’t do math”)

If RepresentationHitRate and RetestGain_7d are low, the child didn’t “fail math”—the repair loop failed.

Failure Trace (schematic)

Weak representation → wrong method → chain breaks → time loss → panic → TTC collapse → avoidance (NIT) → failure

Repair Corridor (APRC): How to fix it without overload

Detect which bind broke: representation? method choice? steps? checking? time?
Truncate (stop full papers temporarily): micro drills 10–15 min.
Preserve core continuity: keep 1 weekly “win anchor” (confidence buffer).
Stitch micro → mixed mini-set → timed section → full paper.
Widen corridor only after TTC improves and variance drops.

Parent Actions (simple, high leverage)

Stop asking “do more papers” first. Ask: “Which mistake repeated?”
Enforce correction + retest within 7 days for recurring errors.
Train one universal word-problem routine:

underline what is asked,
list given numbers with units,
draw model,
choose operation,
check reasonableness.

Convert “careless” into checklists (units, signs, reread question, estimate).
Keep timed practice small until TTC rises.

InterstellarCore (optional P3 seed — small but powerful)

Once P2 is stable, add 1 transfer task/week:

explain a solution in 3 sentences,
show a second method,
create a similar question and solve it,
debug a wrong solution (find first wrong step).

Purpose: build transfer + self-correction without destabilising the pass corridor.

Almost-Code Spec Block (copy-paste)

“`text id=”p_math_fail_v12″
SPEC: EducationOS.PrimaryMath.FailureAnalysis.Works.v1.2

ENTITY:
Learner.Z0
ParentOrTutor.Z1
SchoolTest.Z2

CHRONOFLIGHT:
State = {Z, P, Load, Drift, Repair, Buffer, Transfer, Coupling}
RouteStates = {Climbing, StableCruise, Drift, CorrectiveTurn, Descent}
TimeSlice = 1 week
Objective = move P0/P1 -> P2 stable; seed P3 transfer (optional)

PHASE_MAP:
P0: wrong operations + chain breaks; TTC low; blanks; NIT triggers
P1: routine works; variation breaks
P2: stable pass corridor; correct modeling + method selection; completes in time
P3: transfer corridor; explain + alternate method + self-correct

RATE_DOMINANCE_LAW:
FAIL if Ğ(repair+correction+retest+method_training+timed_stability) < Ḋ(drift+confusion+time_pressure+careless_cascade) PASS if Ğ >= Ḋ for >= 4-8 weeks

KERNEL_NODES(Z0):
ArithmeticFluency
NumberSense
RepresentationModeling
MethodSelection
MultiStepStability
CheckingRecovery
TimeExecution
MisconceptionControl

CRITICAL_BINDS:
Language->Representation
Representation->Method
Method->Steps
Answer->Check->Recover
Error->Correction->Retest

FENCEOS_SENSORS:
TTC
RepresentationHitRate
MethodSelectionRate
StepErrorRate
CarelessErrorRate
CompletionRate
RetestGain_7d
NIT

FAILURE_TRACE:
RepresentationBreak -> WrongMethod -> ChainBreak -> TimeLoss -> TTC_Collapse -> NIT -> Descent

REPAIR_CORRIDOR(APRC):
Detect(bind-break) ->
Truncate(micro drills) ->
PreserveCoreWin(weekly anchor) ->
Stitch(micro->mixed->timed section->paper) ->
Retest(within 7 days) ->
WidenVariation(only when TTC up, variance down)

INTERSTELLARCORE_SEED(optional):
1 TransferTask/week:
{explain in 3 sentences, second method, create variant, debug first wrong step}

END SPEC
“`

Why My Child Failed Primary Mathematics (ChronoFlight × CivOS/FenceOS — Parent-Tailored)