[SITE]
Domain: edukatepunggol.com
Place:  SGP.SG.PG
Lane:   EDU.REPAIR (learning recovery + upgrade)
Method: FENCING (controlled engagement under load)
Scope:  Primary 1 → University
Zoom:   Z0–Z3
Spec:   Unified v1.0

[NODE]
SGP.SG.PG×EDU×Z3×V×LOCK×EKPG.FENCING.MASTER×v1.0:
  txt: "Fencing Method is a universal learning control system: contain load, isolate failure, repair primitives, then expand variation/time—speed last—so students move P0→P3 across repeated S-curves."

[NODE]
SGP.SG.PG×EDU×Z3×S×DEF×FENCING.DEF×v1.0:
  def: "Fencing = deliberate containment of learning difficulty so errors remain isolated, repairable, and non-destructive to stability/identity; the fence expands only after stability is proven."

[NODE]
SGP.SG.PG×EDU×Z3×S×PHASE×P0P3.FLIGHT×v1.0:
  P3: "robust under novelty + time; self-corrects; stable output"
  P2: "routine competent; brittle under variation/time"
  P1: "high variance; depends on hints; confidence fragile"
  P0: "shutdown/avoidance/blanking; identity damage risk"

[NODE]
SGP.SG.PG×EDU×Z3×S×STACK×S0S7.UNIVERSAL×v1.0:
  S0 INPUT/DECODING:
    - "attention, precision reading, symbol decoding"
  S1 REPRESENTATION/MODELS:
    - "mental models; words→structure; meaning-first"
  S2 RETENTION:
    - "active recall; spaced reinforcement; decay prevention"
  S3 ERROR LOOP:
    - "detect→localise→correct; repeat-error elimination"
  S4 TRANSFER:
    - "handle variation; twist questions; method selection"
  S5 META-CONTROL/TIME:
    - "planning; sequencing; time budgeting; skip/return"
  S6 STABILITY:
    - "stress regulation; recovery after slip; no panic/blanking"
  S7 MEANING/IDENTITY:
    - "persistence; self-belief; goals without shame"

[NODE]
SGP.SG.PG×EDU×Z3×S×MODEL×LEARNING.SCURVE×v1.0:
  phases:
    - "slow start: decoding + representation fragile"
    - "rapid climb: automation + pattern fluency"
    - "plateau: stable performance (low variance)"
  collapse_rule:
    - "if load > stability band during slow start → drift P2→P1→P0"
  fencing_role:
    - "shrink fence during slow start so success frequency remains high"

[NODE]
SGP.SG.PG×EDU×Z3×S×MODEL×SKILL.NETWORK.METCALFE×v1.0:
  mapping:
    - "nodes = mastered primitives/concepts"
    - "edges = connections (transfer links, cues, methods, proofs)"
  implication:
    - "value grows superlinearly as nodes+edges increase"
  result:
    - "students who stay across years build dense networks → faster retrieval, stronger transfer, lower panic probability, stable P3 default"

[LAWS:FENCING]
LAW_1 CONTAINMENT:
  - "failure must be contained before repair is possible"
LAW_2 LOAD_LIMIT:
  - "learning works only inside a stability band"
LAW_3 SUCCESS_FREQUENCY:
  - "success frequency must exceed failure frequency during repair"
LAW_4 ONE_VARIABLE:
  - "change only one variable at a time (difficulty OR variation OR time)"
LAW_5 SPEED_LAST:
  - "introduce time, then speed only after stability"

[NODE]
SGP.SG.PG×EDU×Z3×R×PROTOCOL×FENCING.LOOP×v1.0:
  1 TRUNCATE:
    - "reduce scope immediately"
    - "remove time pressure"
    - "stop mixed sets/full papers"
    - "stop volume escalation"
  2 ISOLATE:
    - "identify failing primitive (S0..S7)"
  3 REPAIR_INSIDE_FENCE:
    - "high success frequency"
    - "explicit error localisation scripts"
    - "repeat-error elimination"
  4 EXPAND_VARIATION:
    - "controlled variations (same principle, new surface)"
  5 ADD_TIME:
    - "time budgeting; mild time constraints"
  6 SPEED_LAST:
    - "increase pace only when variance stays low"
  output:
    - "P1→P2→P3 stable flight"

[NODE]
SGP.SG.PG×EDU×Z3×S×LAW×FENCING.UNIVERSALITY×v1.0:
  reason:
    - "all subjects require: input → representation → retrieval → correction → transfer → time control → stability"
  therefore:
    - "fencing controls the learning system, not the subject content"

[TRAJECTORY:PRIMARY_TO_UNI]
Primary 1–2:
  new_scurves: decoding, attention, basic number sense
  fence_focus: S0 + S1 + S2
Primary 3–4:
  new_scurves: multi-step reasoning, explanation, comprehension depth
  fence_focus: S1 + S3
Primary 5–6 (PSLE mode):
  new_scurves: transfer + time compression
  fence_focus: S4 + S5 + S6
Secondary 1–2:
  new_scurves: algebra decoding, new marking logic, higher abstraction
  fence_focus: S0 + S1 + S3
Secondary 3–4 (SEC mode):
  new_scurves: mixed papers, long chains, exam stability
  fence_focus: S4 + S5 + S6
JC/IB/Poly:
  new_scurves: abstraction, proof-like reasoning, higher concurrency
  fence_focus: S1 + S4 + S5 + S6
University:
  new_scurves: autonomy, multi-module coordination, deep transfer
  fence_focus: S5 + S6 (coordination + stability) + S4 (transfer)

[ADAPTER]
SGP.SG.PG×EDU.FENCING×Z2×S×ADAPTER×MATH×v1.0:
  typical_failures:
    - "wrong method selection (S4)"
    - "long-chain error bleed (S3)"
    - "time end-collapse (S5/S6)"
  fencing_moves:
    - "short chains first"
    - "checkpointing + fast localisation"
    - "variation ladder"
    - "time last; speed last"

[ADAPTER]
SGP.SG.PG×EDU.FENCING×Z2×S×ADAPTER×ENGLISH×v1.0:
  typical_failures:
    - "misread questions (S0)"
    - "weak structure (S1/S5)"
    - "rambling under time (S6)"
  fencing_moves:
    - "question parsing rituals"
    - "passage/argument mapping"
    - "paragraph intent templates"
    - "time last; speed last"

[ADAPTER]
SGP.SG.PG×EDU.FENCING×Z2×S×ADAPTER×SCIENCE×v1.0:
  typical_failures:
    - "memorisation without causal model (S1)"
    - "application fails in new context (S4)"
  fencing_moves:
    - "concept models (cause→effect chains)"
    - "keyword precision"
    - "new-context ladder"
    - "time last; speed last"

[ADAPTER]
SGP.SG.PG×EDU.FENCING×Z2×S×ADAPTER×HUMANITIES×v1.0:
  typical_failures:
    - "content dumping without argument (S1/S5)"
    - "weak evidence selection (S4)"
  fencing_moves:
    - "claim→evidence→explain templates"
    - "evidence banks mapped to claims"
    - "compare/evaluate ladder"
    - "time last; speed last"

[ADAPTER]
SGP.SG.PG×EDU.FENCING×Z2×S×ADAPTER×PROGRAMMING×v1.0:
  typical_failures:
    - "copying code without model (S1)"
    - "panic debugging (S3/S6)"
  fencing_moves:
    - "small functions; predictable I/O"
    - "debug loop: reproduce→isolate→hypothesis→test→fix"
    - "edge-cases ladder"
    - "time last; speed last"

[NODE]
SGP.SG.PG×EDU×Z3×R×TRACE×FENCING.FAILURE_TRACE×v1.0:
  unfenced:
    - "load jump"
    - "student fails early"
    - "system increases volume/speed"
    - "failure frequency rises"
    - "identity damage"
    - "P2→P1→P0"
  fenced:
    - "load jump"
    - "truncate + isolate"
    - "repair inside fence (high success)"
    - "expand variation"
    - "add time"
    - "speed last"
    - "P1→P2→P3"

[NODE]
SGP.SG.PG×EDU×Z3×V×LOCK×FENCING.LONG_STAY.CYCLES×v1.0:
  txt: "Students who stay from Primary to Secondary (and beyond) complete multiple fenced S-curves. Each cycle adds nodes and edges (Metcalfe), lowering collapse risk and making P3 stability the default."

[NODE]
SGP.SG.PG×EDU×Z3×R×MODULE×EKPG.FENCING.OPERATIONS×v1.0:
  rules:
    - "one variable at a time"
    - "success frequency > failure frequency"
    - "repeat-error elimination before speed"
    - "time introduced only after clarity"
    - "students learn repair scripts (self-correction)"

[LOCKBLOCK:EDUKATE_PUNGGOL.FENCING.UNIVERSAL.v1.0]
Fencing Method works for every subject because it controls the learning system, not the content: contain load during slow-start S-curves, isolate failing primitives (S0–S7), repair with high success frequency, then expand variation and time—speed last.
Over Primary→Secondary→Uni, repeated fenced cycles build a dense skill network (Metcalfe), making P3 stability the default.

[INTERNAL_LINKS:SPINE]
- Full Stack Education OS (Primary→PSLE→Secondary→SEC + Full SBB)
- Exam Sensors (PSLE + SEC)
- Parent Protocol (load + stability)
- English Lane Overlay
- Primary Math/Science Overlays
- Secondary E-Math Overlay
- Additional Mathematics Overlay

[EXTENSION]
Base_Method: FENCING
Domain: edukatepunggol.com
Scope:  University + Professional Exams
Zoom:   Z2–Z4 (modules → programs → professions)
Spec:   Unified v1.0 (inherits FENCING.UNIVERSAL)

[NODE]
SGP.SG.PG×EDU×Z4×V×LOCK×FENCING.UNI_PRO.MASTER×v1.0:
  txt: "At University and professional levels, failure is dominated by autonomy, concurrency, and stakes. Fencing controls coordination load, isolates failure at module/competency granularity, and restores P3 stability under high pressure."

[NODE]
SGP.SG.PG×EDU×Z4×S×DELTA×UNI_PRO.CHANGES×v1.0:
  changes:
    - "high autonomy (self-scheduling)"
    - "multi-module concurrency"
    - "abstract representations"
    - "sparse feedback"
    - "high-stakes, infrequent exams"
  invariants:
    - "S0–S7 skill stack"
    - "S-curve acquisition"
    - "Metcalfe compounding"
    - "P0–P3 phase physics"

[NODE]
SGP.SG.PG×EDU×Z4×S×PRIM×UNI_PRO.LOAD_TYPES×v1.0:
  L1 CONTENT_LOAD:
    - "concept density, abstraction"
  L2 COORDINATION_LOAD:
    - "multiple modules/tasks in parallel"
  L3 TIME_HORIZON_LOAD:
    - "long semesters, delayed exams"
  L4 STAKES_LOAD:
    - "career/licensing consequences"
  L5 IDENTITY_LOAD:
    - "professional self-image risk"

[NODE]
SGP.SG.PG×EDU×Z4×S×PHASE×P0P3.UNI_PRO×v1.0:
  P3:
    - "stable under autonomy + concurrency"
    - "self-repairs without supervision"
    - "performs under stakes"
  P2:
    - "competent but workload brittle"
    - "peaks near exams only"
  P1:
    - "procrastination + cramming cycles"
    - "panic near deadlines"
  P0:
    - "burnout, avoidance, exam failure"

[NODE]
SGP.SG.PG×EDU×Z4×S×MODEL×UNI.NESTED_SCURVES×v1.0:
  curves:
    - "topic-level S-curves"
    - "module-level S-curves"
    - "semester-level S-curves"
  collapse_rule:
    - "early topic failure propagates upward if unfenced"
  fencing_role:
    - "contain at lowest curve possible (topic), not semester"

[NODE]
SGP.SG.PG×EDU×Z4×S×CONTROL×FENCE.DIALS×v1.0:
  D1 SCOPE:
    - "topic / subtopic / competency"
  D2 STEPS:
    - "number of reasoning steps allowed"
  D3 NOVELTY:
    - "known pattern → slight variation → unseen"
  D4 TIME:
    - "no time → mild → exam time"
  D5 STAKES:
    - "practice → mock → licensure-level"
  rule:
    - "adjust only one dial at a time during repair"

[NODE]
SGP.SG.PG×EDU×Z4×R×PROTOCOL×FENCING.UNI.LOOP×v1.0:
  1 TRUNCATE:
    - "freeze new topics temporarily"
    - "reduce concurrent modules if possible"
    - "remove timed mocks"
  2 ISOLATE:
    - "select one topic within one module"
    - "map failure to S0–S7"
  3 REPAIR:
    - "high success frequency tasks"
    - "explicit self-check scripts"
  4 STITCH:
    - "reconnect topic to module map"
  5 EXPAND:
    - "add variation, then time"
  6 COORDINATE:
    - "reintroduce next module"

[NODE]
SGP.SG.PG×EDU×Z4×S×MODEL×PRO.EXAM.MODEL×v1.0:
  characteristics:
    - "low frequency, high stakes"
    - "broad syllabus"
    - "time compression"
    - "application-heavy"
  dominant_failures:
    - "content sprawl"
    - "weak transfer"
    - "panic under stakes"

[NODE]
SGP.SG.PG×EDU×Z4×R×PROTOCOL×FENCING.PRO.EXAM×v1.0:
  TRUNCATE:
    - "narrow to tested competencies"
    - "suspend full-length mocks initially"
  ISOLATE:
    - "one competency × one question archetype"
  REPAIR:
    - "decision trees"
    - "error-pattern elimination"
  STITCH:
    - "competency → domain → full paper"
  TIME_LAST:
    - "section pacing"
    - "end-game stability drills"

[ADAPTER]
DOMAIN: MEDICINE
Failures:
  - "memorisation without mechanism (S1)"
  - "case-transfer failure (S4)"
  - "panic in long exams (S6)"
Fencing:
  - "mechanism maps (pathophysiology chains)"
  - "case archetypes"
  - "timed blocks last"

[ADAPTER]
DOMAIN: LAW
Failures:
  - "issue-spotting misses (S0/S1)"
  - "argument structure collapse (S5)"
Fencing:
  - "issue trees"
  - "IRAC templates"
  - "essay timing last"

[ADAPTER]
DOMAIN: ENGINEERING
Failures:
  - "formula dumping without system model (S1)"
  - "multi-step error bleed (S3)"
Fencing:
  - "system diagrams"
  - "checkpointed derivations"
  - "exam speed last"

[ADAPTER]
DOMAIN: FINANCE
Failures:
  - "concept overload (S1)"
  - "time misallocation (S5)"
Fencing:
  - "decision flowcharts"
  - "question-type clustering"
  - "section pacing drills"

[ADAPTER]
DOMAIN: TECH
Failures:
  - "copy-paste knowledge (S1)"
  - "debug panic (S3/S6)"
Fencing:
  - "small-function drills"
  - "debug hypothesis loops"
  - "timed coding last"

[NODE]
SGP.SG.PG×EDU×Z4×S×LAW×PRO.METCALFE×v1.0:
  txt: "At professional level, dense competency networks reduce marginal study time per topic and sharply lower exam volatility."

[NODE]
SGP.SG.PG×EDU×Z4×R×TRACE×UNI_PRO.FAILURE×v1.0:
  unfenced:
    - "autonomy overload"
    - "content sprawl"
    - "late cramming"
    - "panic under stakes"
    - "P1→P0"
  fenced:
    - "truncate scope"
    - "competency isolation"
    - "repair inside fence"
    - "expand carefully"
    - "time last"
    - "P2→P3"

[NODE]
SGP.SG.PG×EDU×Z4×V×LOCK×UNI_PRO.OUTCOME×v1.0:
  txt: "Success means stable performance under autonomy, concurrency, and stakes—not peak cramming scores."

[LOCKBLOCK:EDUKATE_PUNGGOL.FENCING.UNI_PRO.v1.0]
Fencing Method scales to University and professional exams by controlling autonomy and concurrency load: isolate competencies, repair inside narrow fences, then expand variation and time—speed last.
Nested S-curves and Metcalfe compounding make P3 stability the default even under high stakes.

[NODE]
SGP.SG.PG×EDU×Z3×S×DEF×FENCE.SIZE×v1.0:
  def: "Fence size is the total effective load applied to a learner during practice or assessment. It is controlled by five independent dials."

[DIAL]
ID: D1.SCOPE
Controls:
  - number of topics
  - number of concepts
  - number of question types
States:
  narrow → moderate → wide
Rule:
  - shrink scope first when failure appears

[DIAL]
ID: D2.STEPS
Controls:
  - number of reasoning steps
  - chain length
  - dependency depth
States:
  single-step → multi-step → long-chain
Rule:
  - reduce steps before reducing content

[DIAL]
ID: D3.NOVELTY
Controls:
  - familiarity of problem surface
  - variation level
States:
  known pattern → slight variation → unseen
Rule:
  - novelty added only after correctness stabilises

[DIAL]
ID: D4.TIME
Controls:
  - time pressure
  - pacing constraint
States:
  untimed → soft timing → exam timing
Rule:
  - time is introduced AFTER accuracy and transfer

[DIAL]
ID: D5.STAKES
Controls:
  - emotional consequence
  - perceived cost of failure
States:
  low (practice) → medium (mock) → high (real exam)
Rule:
  - never combine high stakes with repair

[LAW]
SGP.SG.PG×EDU×Z3×S×LAW×FENCE.ADJUSTMENT×v1.0:
  rule:
    - "Adjust only ONE dial at a time."
  priority_order:
    - D1 Scope
    - D2 Steps
    - D3 Novelty
    - D4 Time
    - D5 Stakes

[ROUTING]
If errors are random:
  action: shrink D2.STEPS
If student cannot start:
  action: shrink D1.SCOPE + D2.STEPS
If student succeeds in drills but fails twists:
  action: shrink D3.NOVELTY
If end-of-paper collapse:
  action: shrink D4.TIME
If panic/avoidance appears:
  action: shrink D5.STAKES immediately

[LAW]
SGP.SG.PG×EDU×Z3×S×LAW×FENCE.EXPANSION×v1.0:
  expand_only_if:
    - success_rate ≥ 80%
    - repeat_error_rate = 0
    - emotional stability maintained
  expansion_order:
    - D3 Novelty
    - D4 Time
    - D5 Stakes

[LOCKBLOCK:FENCE.SIZE.CONTROLS.v1.0]
Fence size is controlled by five dials: scope, steps, novelty, time, stakes.
Repair requires shrinking the fence; acceleration requires expanding it.
Only one dial moves at a time. Speed is always last.

[NODE]
SGP.SG.PG×EDU×Z4×S×PROBLEM×UNI.AUTONOMY.COLLAPSE×v1.0:
  cause:
    - "high autonomy"
    - "multi-module concurrency"
    - "delayed feedback"
    - "high stakes"
  result:
    - "coordination overload"
    - "procrastination loops"
    - "cramming collapse (P1→P0)"

[NODE]
SGP.SG.PG×EDU×Z4×S×PRIM×COORDINATION.LOAD×v1.0:
  def: "load created by managing multiple learning streams in parallel"
  note:
    - "dominant failure cause at university level"

[NODE]
SGP.SG.PG×EDU×Z4×S×DEF×AUTONOMY.FENCE×v1.0:
  def: "An autonomy fence limits how many modules, topics, and deadlines a learner actively engages with at once."

[NODE]
SGP.SG.PG×EDU×Z4×R×PROTOCOL×UNI.FENCE.CONSTRUCTION×v1.0:
  steps:
    1) "select ONE primary module"
    2) "select ONE core topic within that module"
    3) "freeze all other modules at maintenance mode"
    4) "apply standard FENCING.LOOP to the topic"

[NODE]
SGP.SG.PG×EDU×Z4×S×MODE×MAINTENANCE.MODE×v1.0:
  definition:
    - "minimum effort to prevent decay without expansion"
  actions:
    - "light review"
    - "no new topics"
    - "no timed mocks"

[SCHEDULER]
SGP.SG.PG×EDU×Z4×R×MODULE×UNI.WEEKLY.FENCE×v1.0:
  week_structure:
    - 70% effort → fenced primary module
    - 30% effort → maintenance modules
  rule:
    - "only one module allowed expansion per week"

[NODE]
SGP.SG.PG×EDU×Z4×R×PROTOCOL×UNI.EXAM.FENCE×v1.0:
  actions:
    - "stop new content globally"
    - "narrow to tested competencies"
    - "fence to question archetypes"
    - "time introduced last"

[VARIANT]
DOMAIN: PROFESSIONAL_EXAMS
Fence:
  - "competency-based fencing"
  - "decision-tree isolation"
  - "section-by-section expansion"
Key rule:
  - "never run full-length mocks during repair"

[TRACE]
unfenced:
  - "study everything lightly"
  - "nothing stabilises"
  - "panic near exam"
  - "collapse"
fenced:
  - "one module stabilised"
  - "transfer restored"
  - "time added late"
  - "controlled exam performance"

[LOCKBLOCK:UNI.AUTONOMY.FENCE.v1.0]
University failure is usually coordination overload, not intelligence.
Autonomy fencing limits concurrent expansion: one module grows, others maintain.
Stability compounds; collapse probability drops.