SG Exam Simulation OS v0.1 — Timed Load Simulator + Phase Reliability Dashboard (Primary → Secondary → JC)

TITLE: SG Exam Simulation OS v0.1 — Timed Load Simulator + Phase Reliability Dashboard (Primary → Secondary → JC)
VERSION: CivOS Unified Spec v1.x
MODE: Almost-Code / LLM-first / WordPress paste-ready
OWNER: eduKateSG Fence Learning Systems
SCOPE: Singapore exam performance as a controlled load environment across subjects
GOAL: Predict collapse before it happens, fence early, stitch fast, and build P2/P3 stability
CANONICAL CLAIM:
- Exams are not “tests of knowledge”. They are controlled load spikes.
- Most failures are Phase failures (P2→P0) caused by:
  prerequisite debt + low buffer + high coupling + late fencing.
- A simulator makes collapse predictable and repair routable.
========================================================
0) GLOBAL PRIMITIVES (LOCKED)
========================================================
Phase: P0..P3 reliability under load
Zoom: Z0 learner (primary), extends to Z1/Z2 via reporting
FenceOS: R, Θ, Λ
APRC: Truncation + Stitching
Coupling: Cross-subject dependency weights
NIT: English/Humanities; NIT-S: Science mechanisms
========================================================
1) SYSTEM DEFINITION
========================================================
System: SG-EXAM-SIM-OS
Type: Load Simulator + Diagnostic Router
Inputs: performance traces under timed conditions
Outputs: Phase score, collapse risk, repair route, weekly CH/ai plan
Exam := a standardized load spike applied to multiple coupled nodes in limited time.
========================================================
2) WHAT THIS OS ADDS (BEYOND SUBJECT SYLLABUS)
========================================================
Subject lattices tell you “what nodes exist”.
Exam Simulation OS tells you:
- which nodes collapse under timing,
- which coupling edge caused it,
- whether failure is speed, meaning, mechanism, or coherence,
- what to truncate now,
- what to stitch first,
- when to re-test.
========================================================
3) INPUTS (DATA YOU COLLECT)
========================================================
Per Subject, collect:
Math:
- TT1 (timed core node set)
- TT-M (mixed topics)
- SII-math (speed stability)
- TR-math (transfer reliability)
English:
- RR (retrieval reliability)
- GR (grammar reliability timed)
- CR (coherence reliability)
- SR (structure reliability)
- NIT triggers count
- timed composition output rate (words/min)
Science:
- DR (definition reliability)
- MR (mechanism reliability)
- GRF (graph/formula reliability)
- PR (practical evaluation reliability)
- NIT-S triggers count
Humanities:
- TCR (timeline coherence)
- CR (causality reliability)
- ER (evidence reliability)
- SR (structure reliability)
- NIT-ARG triggers count
Global:
- TTC (weeks)
- BN (backlog nodes)
- HoursPerWeek (buffer)
- VolatilityIndex (CCA/events/illness)
- CouplingRiskScore (from coupling table)
========================================================
4) OUTPUTS (WHAT THE SIMULATOR PRODUCES)
========================================================
OutputRecord := {
  PhaseScorePerSubject,
  CollapseRiskScore,
  DominantFailureMode,
  CouplingCause (if any),
  TruncationPlan,
  StitchPlan,
  ReTestSchedule,
  CHaiWeeklyRoutePlan
}
========================================================
5) PHASE SCORING (HOW WE ASSIGN P0–P3)
========================================================
Per Subject, compute Phase by timed transfer behavior.
P0:
- cannot complete / random answers / severe coherence/mechanism collapse
P1:
- succeeds with scaffolding OR in drills, fails in timed transfer
- high VFR, low TR
P2:
- reliable on typical exam mix; minor errors; stable completion
P3:
- stable under high variation; strong speed control; recovers from surprises
PhaseScore := weighted function of:
- completion rate
- TR
- error clustering
- NIT/NIT-S trigger frequency
- speed stability
========================================================
6) COLLAPSE RISK MODEL (CivOS-HD)
========================================================
CollapseRiskScore (CRS) increases with:
- Low Buffer: HoursPerWeek low, sleep/time thin
- High CouplingRiskScore: upstream deficits driving downstream failure
- High BacklogNodes (BN)
- High DriftVelocity: d(TR)/dt falling week-on-week
- High NIT frequency
- TTC short
FenceOS-inspired qualitative thresholds:
- If TTC < 8 weeks AND BN > 3 → Emergency Mode
- If NIT triggers ≥2 per week → Coherence/Mechanism Repair Mode
- If TR < 70% on mixed tests → Transfer Repair Mode
========================================================
7) DOMINANT FAILURE MODE CLASSIFIER (THE “WHY”)
========================================================
Classify failure into one dominant mode:
FM-1 SpeedCollapse:
- knows method but cannot finish; SII high; accuracy drops under time
FM-2 RetrievalCollapse:
- cannot pull vocab/formulas under time; RR/GRF low
FM-3 Meaning/CoherenceCollapse (NIT):
- English/Humanities structure breaks; contradictions appear
FM-4 MechanismCollapse (NIT-S):
- Science explanations wrong; formula plugging without causal chain
FM-5 CouplingMisrepair:
- repairing the wrong subject; upstream deficit remains
FM-6 BufferHollowing:
- too many topics; no time to stitch; Θ≥1 scenario
========================================================
8) TRUNCATION PLAN (WHAT TO DROP NOW)
========================================================
TruncationPlan rules (default):
If Emergency Mode:
- Drop low-weight enrichment nodes
- Focus only on VH/H coupling edges + core nodes per subject
- Reduce writing complexity (English/Hum) to coherence-first templates
- Reduce science breadth to core mechanism spines + data/practical
- Reduce math breadth to prereq-critical chain + WP structure
If SpeedCollapse:
- shorten tasks, ramp timing gradually
- enforce method selection and “checkpoint” habits
If NIT / NIT-S:
- truncate complexity (claims/plot/mechanism length)
- rebuild 2-link chain stability first
========================================================
9) STITCH PLAN (WHAT TO REBUILD FIRST)
========================================================
StitchPlan priority order:
Priority-1: Upstream VH Couplings
- ENG inference → MATH WP
- MATH graphs/algebra → PHY motion/electricity
- SCI data graphs → GEO data
- ENG paragraph → HUM argument coherence
Priority-2: Core node stability (per subject triage list)
- Math: FRAC/RPP/WP/ALG
- English: READ-INF/VOC/GRAM/PARA/PLAN
- Science: DATA/PRAC + PHY-FORCES/MOTION + CHE-PARTICLES/REACTIONS + BIO-CELLS/TRANSPORT
- Humanities: TIMELINE/CAUSE/EVIDENCE/THESIS
Priority-3: Variation and mixed transfer training
- TT-V and TT-M after stability returns
========================================================
10) RE-TEST SCHEDULE (THE FEEDBACK LOOP)
========================================================
ReTestSchedule:
Weekly:
- Math TT1 + small TT-M
- English RR test + 1 timed paragraph + NIT check
- Science DR + 90s mechanism explain + GRF mini-test
- Humanities 1 causality paragraph + evidence check + NIT-ARG check
Biweekly:
- mixed cross-subject micro-sim (mini exam block)
Monthly:
- full exam simulation (only after P2 stable on core nodes)
========================================================
11) CH/ai WEEKLY ROUTE PLAN GENERATOR (SIMULATOR OUTPUT)
========================================================
CH/ai.Inputs:
- PhaseScorePerSubject
- CollapseRiskScore
- DominantFailureMode
- CouplingCause
- TTC, HoursPerWeek, VolatilityIndex
CH/ai.Output:
- WeeklyRoutePlan = {RepairSessions, StabilitySessions, LoadSimSessions, BufferSessions}
Default WeeklyRoutePlan (normal mode):
- 3 Repair Sessions (highest-priority nodes)
- 2 Stability Sessions (mixed practice)
- 1 Load Simulation Session (timed)
- 1 Buffer Session (stitch + catch-up)
Emergency WeeklyRoutePlan:
- 4 Repair Sessions (upstream couplings first)
- 1 Stability Session
- 1 Timed micro-sim (short)
- 1 Buffer rebuild session (sleep/routine + backlog reduction)
========================================================
12) DASHBOARD (WHAT YOU SHOW PARENTS/STUDENTS)
========================================================
Dashboard v0.1 (simple):
- Phase per subject: {Math P?, English P?, Science P?, Hum P?}
- CollapseRiskScore: Low / Medium / High
- DominantFailureMode: FM-x
- Top 3 Repair Nodes this week
- 1 Truncation rule in effect (what we are not doing)
- Next Re-Test date
========================================================
13) REQUIRED SHORT FAILURE MODE TRACE (CANONICAL)
========================================================
Trace:
Backlog grows → coupling failures hidden → TTC short → late fencing (Θ≥1) →
NIT/NIT-S triggers → speed collapse → multi-subject P2→P0 exam shock
Fence Repair:
Run micro-sims early → detect dominant failure mode →
truncate breadth → stitch upstream couplings + core nodes →
timed ramps → stabilize P2/P3
========================================================
END: SG Exam Simulation OS v0.1
========================================================

WORDPRESS PASTE-READY PAGE STRUCTURE
H1: SG Exam Simulation OS — Why Exams Cause Collapse (and how to prevent it)
H2: Exams as Load Spikes (Phase reliability, not “IQ”)
H2: Inputs we measure (Math/English/Science/Humanities sensors)
H2: Phase scoring (P0–P3) for each subject
H2: Collapse Risk model (buffer, TTC, coupling, drift velocity)
H2: Dominant Failure Mode classifier (FM-1 to FM-6)
H2: Truncation Plan (what to drop now)
H2: Stitch Plan (what to rebuild first)
H2: Re-test Schedule (closed loop)
H2: CH/ai Weekly Route Plan Generator
H2: Parent/Student Dashboard v0.1
H2: Failure Trace + Repair Trace