SG-EDU Node Library v0.1 — Core Nodes, Sensors, Repair Loops (Preschool → Primary → Secondary (Full SBB) → Post-Secondary)

TITLE: SG-EDU Node Library v0.1 — Core Nodes, Sensors, Repair Loops (Preschool → Primary → Secondary (Full SBB) → Post-Secondary)
VERSION: CivOS Unified Spec v1.x
MODE: Almost-Code / LLM-first / WordPress paste-ready
OWNER: eduKateSG Fence Learning Systems
PLACE: SG
LANE: EDU
PURPOSE:
- Make Singapore education “executable”: a library of the **minimum high-weight nodes** that determine outcomes.
- Each node includes: prerequisites, outputs, sensors, failure modes, repair loops, fence rules.
SYSTEM ANCHORS (facts):
- NEL Framework shapes preschool curriculum for children aged 4–6.
- PSLE Score ranges 4–32 (sum of ALs across 4 subjects).
- Full SBB: from 2024 Sec 1 cohort, streams removed; Posting Groups 1/2/3; subjects offered at G1/G2/G3.
- SEC: SEAB’s SEC portal lists G1/G2/G3 syllabuses (e.g., 2027 school-candidate syllabuses).
========================================================
0) GLOBAL PRIMITIVES (LOCKED)
========================================================
Phase (Reliability under load): P0/P1/P2/P3
Zoom: Z0 learner, Z1 family, Z2 school, Z3 city, Z4 nation, Z5 civilisation, Z6 supranational
FenceOS:
- Truncation: reduce breadth/complexity early when drift detected
- Stitching: repair weakest upstream node; rebuild buffer; re-test transfer
Core Sensors (cross-subject):
- TTC (time-to-next-gate), BN (backlog nodes), Sleep/Load, AvoidSlope
- Transfer Reliability (TR): stable across unseen prompts/questions
- NIT (coherence irreversibility): English/Humanities writing/argument drift
- NIT-S (mechanism irreversibility): Science causal chain drift
========================================================
1) STAGE GATES + SYSTEM LABELS (REFERENCE)
========================================================
PRESCHOOL:
- Curriculum anchor: NEL Framework (4–6).
PRIMARY:
- Gate: PSLE (AL system; score 4–32).
SECONDARY (FULL SBB):
- Posting Groups: PG1/PG2/PG3 are used for Sec 1 posting and guide initial subject levels options.
- Subjects at levels: G1/G2/G3.
- Gate: SEC (syllabuses at G1/G2/G3).
========================================================
2) NODE RECORD TEMPLATE
========================================================
NodeRecord := {
  ID,
  StageBand,
  Domain,
  Prereq[],
  Outputs[],
  PhaseTarget,
  LoadProfile,
  TransferTests[],
  Sensors[],
  FailureModes[],
  RepairLoops[],
  FenceRules[]
}
ID Grammar (LOCK):
SG×<LANE>×Z0×<ROLE>×Node×<NAME>
========================================================
3) PRESCHOOL (K1/K2) — FOUNDATION BUFFER NODES
========================================================
Node: SG×EDU×Z0×Learner×Node×PRE-LANG-ORAL
StageBand: Preschool (NEL 4–6)
Outputs: PRE-READ-READY, PRE-SELFREG
Sensors: {TalkTurns/day, StoryRetell, AvoidSlope}
FailureModes: {SilentChild, ScreenOnlyInput}
RepairLoops: {daily read-aloud + “tell me what happened”}
FenceRules: {keep it light; routines > worksheets}
Node: SG×EDU×Z0×Learner×Node×PRE-SELFREG
Outputs: PRIMARY-LEARN-HABITS
Sensors: {MeltdownFreq, TransitionTime}
FailureModes: {CannotSit, TaskRefusal}
RepairLoops: {micro-tasks + praise + predictable schedule}
FenceRules: {if distress rises → truncate task length; increase frequency}
Node: SG×EDU×Z0×Learner×Node×PRE-NUM-PATTERN
Outputs: MATH-NUMSENSE-P1
Sensors: {CountAccuracy, PatternSpotting}
RepairLoops: {counting games, grouping, simple patterns}
FenceRules: {no heavy drilling; keep curiosity alive}
========================================================
4) PRIMARY NODE SET (P1–P6) — HIGH-WEIGHT CORE
========================================================
-------------------------
4A) ENGLISH CORE NODES
-------------------------
Node: SG×ENG×Z0×Learner×Node×ENG-READ-INF
Prereq: [PRE-LANG-ORAL]
Outputs: ENG-EVIDENCE, ENG-COMP-PLAN
PhaseTarget: P2 by P4–P5
TransferTests: {unseen passage: “why/how” answers with evidence}
Sensors: {TR, AvoidSlope}
FailureModes: {OpinionAnswer, NoEvidence}
RepairLoops: {underline→quote→explain template}
FenceRules: {if inference fails, stop “more practice papers”; repair evidence skill first}
Node: SG×ENG×Z0×Learner×Node×ENG-VOC-RETRIEVAL
Outputs: ENG-SENT-CORE, ENG-WRITE-DETAIL
Sensors: {RR retrieval rate, mis-use rate}
FailureModes: {LabelOnlyTrap, PassiveOnly}
RepairLoops: {Fence Method: word→phrase→sentence→paragraph}
FenceRules: {if RR<60%: truncate new words; stitch retrieval drills}
Node: SG×ENG×Z0×Learner×Node×ENG-PARA-COHERENCE
Prereq: [ENG-SENT-CORE, ENG-READ-INF]
Outputs: ENG-COMP-DRAFT
Sensors: {CR coherence, NIT-lite}
FailureModes: {DriftMidPara, PointNoSupport}
RepairLoops: {topic→3 supports→closing line; checklist}
FenceRules: {if drift repeats: truncate length; stitch paragraph skeleton}
Node: SG×ENG×Z0×Learner×Node×ENG-COMP-PLAN-5BOX
Outputs: ENG-COMP-DRAFT, ENG-COMP-EDIT
Sensors: {SR structure reliability, SII timing stability}
FailureModes: {NoPlanRush, OverPlanNoWrite}
RepairLoops: {timed 5-box plan → write 1st paragraph immediately}
FenceRules: {reduce plan complexity if SR low}
Node: SG×ENG×Z0×Learner×Node×ENG-NIT-CONTROL
Outputs: P3 writing readiness
Sensors: {NIT triggers/week}
FailureModes: {contradiction, causality breaks, random ending}
RepairLoops: {3-scene spine + rewrite weakest paragraph}
FenceRules: {NIT triggers → truncate plot complexity; stitch coherence first}
-------------------------
4B) MATH CORE NODES
-------------------------
Node: SG×MATH×Z0×Learner×Node×MATH-NUMSENSE
Outputs: FRAC-CONCEPT, WP-BAR
Sensors: {CarelessRate, FactFluency}
FailureModes: {SlowBasics, RandomErrors}
RepairLoops: {short daily drills + error tagging}
FenceRules: {if careless spikes: truncate speed; rebuild accuracy, then ramp}
Node: SG×MATH×Z0×Learner×Node×MATH-FRAC-CONCEPT
Outputs: RATIO-RPP, PERCENT
Sensors: {part-whole consistency}
FailureModes: {fraction as two numbers}
RepairLoops: {bar model → equivalence → compare}
FenceRules: {no rushing to advanced topics if fraction concept unstable}
Node: SG×MATH×Z0×Learner×Node×MATH-WP-BAR
Prereq: [ENG-READ-INF (coupling)]
Outputs: PSLE-WP
Sensors: {TR word problems}
FailureModes: {wrong quantities, wrong model}
RepairLoops: {3-pass read→underline→model→equation}
FenceRules: {if WP fails: check reading inference first (coupling repair)}
-------------------------
4C) SCIENCE CORE NODES
-------------------------
Node: SG×SCI×Z0×Learner×Node×SCI-INQ-VARIABLES
Outputs: SCI-DATA, SCI-CONCLUDE
Sensors: {PR practical reliability}
FailureModes: {IV/DV confusion, missing controls}
RepairLoops: {variable sorting drills + checklist}
FenceRules: {enforce control checklist before marking}
Node: SG×SCI×Z0×Learner×Node×SCI-DATA-READ
Prereq: [MATH-FRAC-CONCEPT (coupling)]
Outputs: SCI-EXPLAIN
Sensors: {GRF graph/table reliability}
FailureModes: {reads graph as picture}
RepairLoops: {axis/units discipline + trend statements}
FenceRules: {units mandatory}
Node: SG×SCI×Z0×Learner×Node×SCI-EXPLAIN-BECAUSE
Outputs: PSLE-SCI-STRUCTANS
Sensors: {MR-lite}
FailureModes: {story no mechanism}
RepairLoops: {evidence→because→therefore template}
FenceRules: {if explanation generic: repair mechanism chain, not “memorise more”}
========================================================
5) SECONDARY NODE SET (Sec1–Sec4/5, Full SBB) — CORE + LEVEL LADDER
========================================================
Full SBB context (posting + subject levels):
- Streams removed from 2024 Sec 1 cohort; PG1/PG2/PG3 posting; subjects at G1/G2/G3.
-------------------------
5A) “LEVEL LADDER” NODE (G1/G2/G3)
-------------------------
Node: SG×EDU×Z0×Learner×Node×SEC-LEVEL-LADDER
Purpose: subject levels as stability tooling (not identity)
Inputs: {current subject level, PhaseHealth, TTC, BN}
Outputs: {level adjustment decision + repair plan}
Sensors: {Phase drift, TTC}
FenceRules:
- If chronic P0 risk → level adjustment + core repair loops
- If stable P2 → gradual upgrade + transfer tests
-------------------------
5B) ENGLISH (Secondary) — evidence + argument + writing stability
-------------------------
Node: SG×ENG×Z0×Learner×Node×ENG-SEC-EVIDENCE
Prereq: [ENG-READ-INF]
Outputs: ENG-SEC-STRUCTANS
Sensors: {TR, evidence use rate}
FailureModes: {no quote, no linkage}
RepairLoops: {quote→explain→link}
FenceRules: {mark requires evidence line}
Node: SG×ENG×Z0×Learner×Node×ENG-SEC-WRITE-COHERENCE
Outputs: ENG-SEC-ARG-BASIC
Sensors: {NIT, CR}
FailureModes: {thesis drift, contradiction}
RepairLoops: {3-claim spine + paragraph skeleton}
FenceRules: {NIT triggers → truncate claims; stitch logic chain}
-------------------------
5C) MATH (Secondary) — algebra + graphs coupling-heavy
-------------------------
Node: SG×MATH×Z0×Learner×Node×MATH-ALG-CORE
Outputs: PHY-MOTION, PHY-ELEC, CHE-RATES
Sensors: {algebra error rate under time}
FailureModes: {rearrangement failure}
RepairLoops: {rearrange + substitute + unit tagging}
FenceRules: {if science fails due to algebra: truncate science breadth; stitch algebra first}
Node: SG×MATH×Z0×Learner×Node×MATH-GRAPH-MEANING
Outputs: PHY-MOTION, CHE-RATES, DATA-INTERP
Sensors: {gradient/intercept meaning accuracy}
FailureModes: {graph-as-picture}
RepairLoops: {graph→story→equation→check loop}
FenceRules: {meaning check before calculation}
-------------------------
5D) SCIENCE (Secondary) — mechanism + data + practical
-------------------------
Node: SG×SCI×Z0×Learner×Node×SCI-DATA-GRAPHS
Prereq: [MATH-GRAPH-MEANING]
Outputs: SCI-SEC-STRUCTANS, PRAC-EVAL
Sensors: {GRF, TR}
FailureModes: {describe not explain}
RepairLoops: {trend→evidence→mechanism}
FenceRules: {one mechanism sentence required}
Node: SG×SCI×Z0×Learner×Node×PHY-MOTION
Prereq: [MATH-ALG-CORE, MATH-GRAPH-MEANING]
Sensors: {VFR, MR}
FailureModes: {formula plugging}
RepairLoops: {diagram→story→equation→check}
FenceRules: {if plugging: truncate calculation; stitch meaning chain}
Node: SG×SCI×Z0×Learner×Node×CHE-PARTICLES
Sensors: {DR, MR}
FailureModes: {model-as-picture-only}
RepairLoops: {particle story explanations}
FenceRules: {require “because particles…” line}
Node: SG×SCI×Z0×Learner×Node×BIO-CELLS-FUNCTION
Sensors: {DR, MR}
FailureModes: {label-only}
RepairLoops: {label→function→because}
FenceRules: {no marks without function statement}
Node: SG×SCI×Z0×Learner×Node×PRAC-EVALUATION
Prereq: [SCI-DATA-GRAPHS]
Sensors: {PR}
FailureModes: {generic “repeat experiment” answers}
RepairLoops: {limitation→effect→fix template}
FenceRules: {evaluate one variable at a time}
-------------------------
5E) SEC EXAM NODE (G1/G2/G3)
-------------------------
Node: SG×EDU×Z0×Learner×Node×SEC-EXAM-GATE
Anchor: SEC syllabuses exist at G1/G2/G3.
Outputs: POSTSEC-ROUTING
Sensors: {TTC, BN, TR across subjects}
FenceRules: {Emergency TTC mode if TTC < 8w or BN > 3}
========================================================
6) POST-SECONDARY READINESS NODE SET (JC / Poly / ITE)
========================================================
Node: SG×EDU×Z0×Learner×Node×POSTSEC-ROUTING
Prereq: [SEC-EXAM-GATE]
Inputs: {interests, PhaseHealth, learning style, TTC, portfolio needs}
Outputs: {JC/MI route OR Poly route OR ITE route + plan}
FenceRules:
- choose lane fit + stability under load
- “prestige chasing” without stability → P0 risk
-------------------------
6A) JC READINESS (high-load academic)
-------------------------
Node: SG×EDU×Z0×Learner×Node×JC-READINESS-P2
Sensors: {P2 across Math/Eng/Science, TR, SII}
FailureModes: {fast overload, hidden prereq debt}
RepairLoops: {core-node patching + timed ramps}
FenceRules: {if stability < P2, patch cores before intensity ramp}
-------------------------
6B) POLY READINESS (applied + projects)
-------------------------
Node: SG×EDU×Z0×Learner×Node×POLY-READINESS-OUTPUT
Sensors: {weekly deliverable reliability, self-management}
FailureModes: {deadline collapse}
RepairLoops: {planner + output routines + chunking}
FenceRules: {protect buffers; avoid chronic overload}
-------------------------
6C) ITE READINESS (skills pipeline + progression)
-------------------------
Node: SG×EDU×Z0×Learner×Node×ITE-READINESS-SKILL
Sensors: {practice consistency, hands-on competence}
FailureModes: {irregular practice}
RepairLoops: {repeatable skill loops + coaching}
FenceRules: {consistency > cramming}
========================================================
7) “EXECUTION PACK” (HOW TO RUN THIS LIBRARY WEEKLY)
========================================================
WeeklyRun := {
  Step1: StageGate + TTC
  Step2: PhaseHealth per core subject
  Step3: Identify BN top 1–3 nodes
  Step4: Apply Fence: Truncate breadth + Stitch BN nodes
  Step5: Run Transfer Tests weekly (unseen sets)
  Step6: Only then increase speed/complexity
}
Minimum Transfer Tests (per week):
- English: 1 unseen inference set + 1 paragraph/comp coherence check (NIT)
- Math: 1 WP set + 1 core skills set
- Science: 1 data/mechanism set + 1 practical evaluation prompt
========================================================
END: SG-EDU Node Library v0.1
========================================================