eduKate Singapore

VIEW THIS AS

Auto mode follows the Route Engine until you choose a viewpoint.

YOU ARE HERE

ROUTE CHECK

CONNECTED TO

WHAT NEXT

Use the canonical route for this room, or HELP if you are unsure.

Next useful routeCanonical hubHELP

CivOS Backtest Pack v1.0

Three students discuss open schoolbooks and handwritten notes during a group study session.

(Almost-Code templates — forces RunState → VerifyLog → ThresholdUpdate → VersionBump)

You can paste any historical event, news story, office conflict, or family issue into Template A, then pick one of the 4 specialised templates depending on the use case.

0) Backtest Rules (Definition Lock)

Backtest (CivOS): run the protocol on information available at t0, produce a RunState, then compare to what actually happened by t1, producing a VerifyLog.

Hard rule: A backtest is invalid if it uses post-t0 info to set Phase/Action.

Closure rule: If TotalScore < 70, you must output a ThresholdUpdate and VersionBump.

1) Universal Backtest Template (A) — Use for everything

CivOS.Backtest.A.Universal.v1:
  Meta:
    CaseID: <string>
    Domain: <conflict|city|education|governance|family|office|news|other>
    Place: <PlaceID or null>
    SourceLinks: [<optional>]
  Time:
    t0: <iso8601>     # "what you knew then"
    t1: <iso8601>     # "what happened later"
    Δt_days: <number>
  InputAt_t0:
    EventStream:
      - timestamp: <iso8601 or null>
        text: <bullet facts known at t0>
        source: <news|conversation|ops log|etc>
    Constraints:
      known_unknowns: [<what you didn't know at t0>]
  RunState:
    # Must use CivOS.RunState.v1 contract
    CivOS.RunState.v1: {}
  PredictionCommitment:
    # To prevent hindsight editing
    WhatWouldIExpectBy_t1:
      - <prediction 1>
      - <prediction 2>
    WhatWouldFalsifyThis:
      - <falsifier 1>
      - <falsifier 2>
  ObservedAt_t1:
    WhatActuallyHappened:
      - <bullet facts>
    OutcomeNotes: <string>
  Verification:
    CivOS.VerifyLog.v1: {}
  Learnings:
    3_Lessons:
      - <lesson 1>
      - <lesson 2>
      - <lesson 3>
    ReusableRuleExtract:
      - <rule fragment to add to spec>

2) Early Detection Template (B) — “did we detect drift early enough?”

Use when the goal is warning before failure.

CivOS.Backtest.B.EarlyDetection.v1:
  Meta: {CaseID: <string>, Domain: <...>, Place: <...>}
  Time: {t0: <...>, t1: <...>, Δt_days: <...>}
  InputAt_t0:
    WeakSignals:
      - <signal 1>
      - <signal 2>
    Noise:
      - <things that distracted observers>
    AssumedPriors:
      coupling: <0..1>
      buffer: <0..1>
  RunState:
    CivOS.RunState.v1: {}
  EarlyWarningDecision:
    WarningLevel: <green|amber|red>
    WarningJustification:
      - <sensor-based reason>
    RequiredActionsIfAmberOrRed:
      - <Fence/Truncate/Stitch/Reroute tasks>
  ObservedAt_t1:
    DidThresholdCross:
      crossed: <yes|no>
      crossing_type: <NIT|RateDominance|BufferCollapse|CouplingCascade|other>
    WhatHappened: [<...>]
  Verification:
    CivOS.VerifyLog.v1: {}
  Calibration:
    IfLateOrMiss:
      which_sensor_failed: [TTC|R|NIT|TVIS|Coupling|ZoomPick|LanePick]
      update: <threshold/routing change>

3) Decision Quality Template (C) — “was the action correct?”

Use when you want to test FenceOS actuation and not just detection.

CivOS.Backtest.C.DecisionQuality.v1:
  Meta: {CaseID: <string>, Domain: <...>, Place: <...>}
  Time: {t0: <...>, t1: <...>, Δt_days: <...>}
  InputAt_t0:
    SituationSummary: <string>
    AvailableLevers:
      Operator: [<levers>]
      Oracle: [<levers>]
      Visionary: [<levers>]
  RunState:
    CivOS.RunState.v1: {}
  ActionPlanCommittedAt_t0:
    PrimaryAction: <Monitor|Fence|Truncate|Stitch|Reroute|Escalate>
    Steps:
      - role: <Operator|Oracle|Visionary>
        task: <string>
        expected_effect: <increase TTC | reduce R | reduce coupling | raise buffer>
    SuccessMetrics:
      - <metric 1>
      - <metric 2>
  ObservedAt_t1:
    WhatWasDoneInReality: [<...>]
    WhatHappened: [<...>]
  Verification:
    CivOS.VerifyLog.v1: {}
  Repair:
    IfWrongAction:
      - propose_alternate_action: <...>
      - why_alternate_wins: <sensor logic>
      - threshold_update: <...>

4) Lane/Zoom Attribution Template (D) — “did we pick the right causal zoom?”

Use when failures come from wrong zoom (too micro vs too macro) or wrong lane (wrong load path).

CivOS.Backtest.D.LaneZoomAttribution.v1:
  Meta: {CaseID: <string>, Domain: <...>, Place: <...>}
  Time: {t0: <...>, t1: <...>, Δt_days: <...>}
  InputAt_t0:
    CandidateZooms: [Z0,Z1,Z2,Z3,Z4,Z5,Z6]
    CandidateLanes: [<LaneA>, <LaneB>, <LaneC>]
    EvidenceByZoom:
      Z0: [<facts>]
      Z2: [<facts>]
      Z4: [<facts>]
  RunState:
    CivOS.RunState.v1: {}
  CompetingHypotheses:
    H1: {Zoom: <Zk>, Lane: <LaneX>, why: <string>}
    H2: {Zoom: <Zk>, Lane: <LaneY>, why: <string>}
  ObservedAt_t1:
    WhichHypothesisMatchedReality: <H1|H2|mixed>
    Evidence: [<...>]
  Verification:
    CivOS.VerifyLog.v1: {}
  Update:
    IfWrongZoomOrLane:
      - add_rule: "When <pattern>, prefer <Zoom/Lane>"
      - add_sensor: <missing measurement>

5) NIT / Narrative Lock Template (E) — “did narrative become irreversible?”

Use when the key failure is meaning lock-in (propaganda loops, identity binding, “no reversal possible”).

CivOS.Backtest.E.NITLock.v1:
  Meta: {CaseID: <string>, Domain: <...>, Place: <...>}
  Time: {t0: <...>, t1: <...>, Δt_days: <...>}
  InputAt_t0:
    NarrativeSnippets:
      - <quote-like paraphrase 1>
      - <quote-like paraphrase 2>
    RepetitionSignals:
      frequency: <low|med|high>
      channels: [<media|leaders|groups|office chat|family>]
    PunishmentOrConstraintSignals:
      - <what happens if dissent occurs>
  RunState:
    CivOS.RunState.v1: {}
  NITAssessmentCommittedAt_t0:
    NIT_value: <0..1>
    regime: <reversible|soft_lock|hard_lock>
    ExpectedNextMovesIfSoftOrHard:
      - <move 1>
      - <move 2>
  ObservedAt_t1:
    DidNarrativeLockTighten:
      yes_no: <yes|no>
      evidence: [<...>]
  Verification:
    CivOS.VerifyLog.v1: {}
  Calibration:
    IfNITMisread:
      - adjust_NIT_threshold: <...>
      - add_indicator: <new signal>

6) Scoring Block (paste into every VerifyLog)

Use exactly this so all backtests are comparable.

CivOS.ScoreBlock.v1:
  CoordScore:
    PhaseCorrect: <0|40>
    ZoomCorrect: <0..30>
    LaneCorrect: <0..30>
    Total: <0..100>
  ActionScore:
    ActionAppropriate: <0..50>
    ActionEffectOnSensors: <0..50>
    Total: <0..100>
  TimingScore:
    TTCMargin: <0..100>
  TotalScore:
    formula: "0.4*Coord + 0.4*Action + 0.2*Timing"
    value: <0..100>
  OutcomeClass: <correct|miss|false_alarm|late|wrong_lane|wrong_zoom|wrong_phase|wrong_action>
  FailureTag: <FT.*>

7) Failure Tags (standard)

CivOS.FailureTags.v1:
  - FT.MISS
  - FT.FALSE_ALARM
  - FT.LATE
  - FT.WRONG_PHASE
  - FT.WRONG_ZOOM
  - FT.WRONG_LANE
  - FT.WRONG_ACTION
  - FT.HIDDEN_COUPLING
  - FT.NIT_MISREAD
  - FT.TVIS_MISREAD
  - FT.TTC_MISREAD
  - FT.RATE_DOMINANCE_MISREAD
  - FT.BUFFER_MISREAD

8) Mandatory Version Bump Rule (closure)

CivOS.UpdateRule.v1:
  if TotalScore.value < 70:
    required:
      - FailureTag must be non-empty
      - ThresholdUpdate.changed = yes
      - specify what_changed ids
      - VersionBump must advance (forward-only)

9) “Start Pack” — which 5 backtests to run first (to lock v1.0)

Run these in order to calibrate different failure modes:

  1. Singapore 1942 (early detection + wrong zoom risk) → Template B or D
  2. Ottoman 1922–23 (attrition + lock-in) → Template B + E
  3. Shivaji corridor warfare episode (decision quality + actuation) → Template C
  4. Empire decline sequence (slow attrition rate dominance) → Template A + B
  5. Modern office/family conflict (CONFLICT.OS generalization) → Template A + E

Backtest #1 — Fall of Singapore (Template B: Early Detection)

This backtest is framed as: could CivOS have issued an amber/red warning early enough given what was already observable when Japan began the Malaya campaign against United Kingdom positions in Malaya?

Time window

  • t0: 1941-12-08 (campaign begins)
  • t1: 1942-02-15 (Singapore surrenders)
  • Δt: 69 days

1) InputAt_t0 (what is “allowed”)

CivOS.Backtest.B.EarlyDetection.v1:
  Meta:
    CaseID: "BT-SG-1942-001"
    Domain: "city|war|conflict"
    Place: "SG"
  Time:
    t0: "1941-12-08"
    t1: "1942-02-15"
    Δt_days: 69
  InputAt_t0:
    WeakSignals:
      - "Attacker momentum appears high; fast operational tempo"
      - "Air/sea control likely decisive for supply + defense"
      - "Forward defense chain vulnerable to rapid collapse"
      - "Overconfidence narrative ('fortress') may be masking buffer reality"
    Noise:
      - "Formal defenses exist; public confidence high"
      - "Prior beliefs about geographic difficulty"
    AssumedPriors:
      coupling: 0.70
      buffer: 0.55

2) RunState (INTENTIONALLY shows what a pre-calibration CivOS might output)

This demonstrates a failure → triggers update + version bump (closure).

CivOS.RunState.v1:
  RunID: "RUN-BT-SG-1942-001"
  Place: "SG"
  Window:
    t0: "1941-12-08"
    t1: "1942-02-15"
    Δt_days: 69
  Coordinates:
    Phase: "P2"                 # <-- WRONG (too optimistic)
    Zoom: "Z2"                  # <-- WRONG (too low)
    Lane: "LANE.city-fortress"  # <-- WRONG lane choice
    RoleFocus: "Mixed"
  Sensors:
    TTC_days: 120               # <-- WRONG (too long)
    Buffer:
      thickness: 0.55           # <-- WRONG (overestimated)
      trend: "flat"
    Drift:
      velocity: 0.35
      acceleration: 0.05
    RateDominance:
      R: 0.90
      regime: "warning"
    NIT:
      value: 0.35
      regime: "reversible"
    TVIS:
      Hope: 0.60
      Grind: 0.55
      Wisdom: 0.30
      imbalance: 0.25
    Coupling:
      cross_lane_weight: 0.60
      contagion_risk: 0.55
  Decision:
    PrimaryAction: "Fence"
    SecondaryActions: ["Monitor"]
    TargetPhase: "P3"
    TargetZoom: "Z2"
    ExecutionRoles:
      Operator: ["OP.harden-local-defense", "OP.hold-lines"]
      Oracle:   ["OR.monitor-changes"]
      Visionary:["VI.none"]
  Confidence:
    CoordConfidence: 0.55
    ActionConfidence: 0.50
    Notes: "Fortress buffers presumed adequate; warning but not critical"

Prediction commitment (what this RunState implies):

  • Expectation: “hold for months / no near-term fracture”
  • Falsifier: “rapid collapse of forward defense chain; TTC collapses below ~30 days”

3) ObservedAt_t1 (what actually happened)

ObservedAt_t1:
  WhatActuallyHappened:
    - "Rapid campaign progression; defenses deteriorated quickly"
    - "Surrender occurred within ~2.5 months"
  OutcomeNotes: "Threshold crossed fast; TTC was much shorter than predicted"

4) Verification (score + failure tag + forced update)

CivOS.VerifyLog.v1:
  RunID: "RUN-BT-SG-1942-001"
  ObservedOutcome:
    happened: "City position collapsed quickly; surrender within 69 days"
    outcome_class: "late"   # detected “some risk” but acted as if not time-critical
  Scores:
    CoordScore: 20          # wrong phase, wrong zoom, wrong lane
    ActionScore: 35         # Fence/Monitor insufficient for regime
    TimingScore: 40         # TTC badly overestimated
    TotalScore: 30
  FailureTag: "FT.WRONG_ZOOM"
  ThresholdUpdate:
    changed: "yes"
    what_changed:
      - "Classifier.Zoom.v1: prefer Z3/Z4 when attacker tempo + air/sea control determines TTC"
      - "Classifier.Lane.v1: shift lane to LANE.air-sea-logistics when supply/control dominates"
      - "Sensor.NIT.v1: treat 'fortress narrative' as soft_lock risk (NIT>=0.4) under load"
      - "Sensor.Buffer.v1: downgrade buffer when coupling+tempo high (buffer trend=down)"
      - "Sensor.TTC.v1: TTC must collapse when coupling high + drift acceleration > threshold"
    why: "Failure caused by fortress narrative inflating buffer and suppressing TTC collapse"
  VersionBump:
    from: "v1.0"
    to: "v1.0.1"

5) What we learn (portable rules)

Lesson 1 — Wrong zoom is the most dangerous error

If you model a war/city-defense problem at Z2 (local fortification) while the true driver is Z3/Z4 (theatre control of air/sea + logistics tempo), you will overestimate buffer and TTC.

Lesson 2 — “Fortress narrative” is an NIT trap

Even when facts are available, narrative can lock perception, preventing the OS from triggering Truncate/Reroute early. In CivOS terms: NIT rising → TTC estimate becomes biased upward unless fenced.

Lesson 3 — Lane must follow the dominant rate equation

If the loss/repair balance is governed by control + supply throughput, the lane is not “city walls”; it is “air/sea/logistics corridor.”

6) Patch excerpt (the reusable rule you keep)

RulePatch.v1.0.1:
  - "If cross_lane_weight >= 0.6 AND attacker tempo high:
       Zoom := Z3 or Z4 (not Z2)
       Lane := corridor-control / logistics-throughput (not fortress-frame)
       Buffer.trend := down
       TTC := min(TTC, 45 days) unless strong counter-signal exists"
  - "If narrative uses absolute claims ('cannot fall', 'impregnable'):
       NIT := max(NIT, 0.45) and require Fence/Truncate consideration"

Backtest #2 — Dissolution of the Ottoman Empire (Template B + E: Early Detection + NIT)

Frame: At t0, could CivOS correctly warn that the imperial regime is beyond repair and that the most probable “stable outcome” is a forced reroute into a new sovereignty structure (Republic), with narrative lock-in accelerating?

Time window

  • t0: 1922-10-11 (Armistice of Mudanya signed) (Wikipedia)
  • t1: 1923-10-29 (Republic proclaimed) (Wikipedia)
  • Δt: 383 days

1) InputAt_t0 (allowed information)

CivOS.Backtest.B.EarlyDetection.v1:
  Meta:
    CaseID: "BT-OTT-1922-001"
    Domain: "conflict|governance|state-transition"
    Place: "ANATOLIA/TR"
  Time:
    t0: "1922-10-11"
    t1: "1923-10-29"
    Δt_days: 383
  InputAt_t0:
    WeakSignals:
      - "War-of-independence trajectory + armistice implies old imperial executive is no longer the control center"
      - "Dual-legitimacy structure: Ankara assembly vs Istanbul imperial authority (legitimacy split)"
      - "External recognition/negotiation is shifting toward the assembly-led government (de-facto sovereignty shift)"
      - "Narrative frame is moving from dynasty/empire → nation/sovereignty (meaning lock risk)"
    Noise:
      - "Imperial symbols and institutions still exist; inertia makes collapse appear slower than it is"
    AssumedPriors:
      coupling: 0.75
      buffer: 0.40

2) RunState (using updated v1.0.1 instincts: zoom-up + corridor/legitimacy lane)

CivOS.RunState.v1:
  RunID: "RUN-BT-OTT-1922-001"
  Place: "ANATOLIA/TR"
  Window:
    t0: "1922-10-11"
    t1: "1923-10-29"
    Δt_days: 383
  Coordinates:
    Phase: "P1"
    Zoom: "Z4"
    Lane: "LANE.legitimacy-sovereignty-transition"
    RoleFocus: "Mixed"
  Sensors:
    TTC_days: 60
    Buffer:
      thickness: 0.30
      trend: "down"
    Drift:
      velocity: 0.70
      acceleration: 0.15
    RateDominance:
      R: 1.10
      regime: "breach"
    NIT:
      value: 0.68
      regime: "soft_lock"
    TVIS:
      Hope: 0.75
      Grind: 0.55
      Wisdom: 0.25
      imbalance: 0.40
    Coupling:
      cross_lane_weight: 0.80
      contagion_risk: 0.75
  Decision:
    PrimaryAction: "Reroute"
    SecondaryActions: ["Truncate", "Stitch", "Fence"]
    TargetPhase: "P2"
    TargetZoom: "Z4"
    ExecutionRoles:
      Operator:
        - "OP.transfer-control-center: align admin, security, finance to new sovereign node"
        - "OP.contain-fragmentation: prevent multi-node legitimacy drift (split command)"
      Oracle:
        - "OR.instrument-legitimacy: track narrative lock, elite defections, external recognition signals"
        - "OR.TTC-refresh: update TTC weekly (legitimacy drift + recognition events)"
      Visionary:
        - "VI.new-constitution-track: define the stable successor structure"
        - "VI.buffer-thicken: create redundancy + continuity rules for state organs"
  Confidence:
    CoordConfidence: 0.78
    ActionConfidence: 0.72
    Notes: "Regime breach is in legitimacy lane; best survival move is controlled transition, not restoration"

PredictionCommitment (what this RunState asserts)

  • By t1, expect formal termination of the imperial sovereignty structure and replacement by a new governance form.
  • Expect NIT tightening (soft_lock → near hard_lock) as “empire” becomes politically non-viable framing.
  • Expect a formal legal milestone sequence (not just military drift): constitutional/assembly acts + treaty normalization.

3) ObservedAt_t1 (what happened)

Key outcomes within the window:

ObservedAt_t1:
  WhatActuallyHappened:
    - "Imperial sovereignty terminated (abolition of sultanate)"
    - "International-legal normalization via treaty"
    - "New republican form formally declared"
  OutcomeNotes: "Transition path matches reroute + truncate + stitch sequence"

4) Verification (score + failure tags + update decision)

CivOS.VerifyLog.v1:
  RunID: "RUN-BT-OTT-1922-001"
  ObservedOutcome:
    happened: "Abolition of sultanate (1922-11-01) → treaty normalization (1923-07-24) → republic proclaimed (1923-10-29)"
    outcome_class: "correct"
  Scores:
    CoordScore: 88
    ActionScore: 80
    TimingScore: 75
    TotalScore: 83
  FailureTag: "FT.NONE"
  ThresholdUpdate:
    changed: "no"
    what_changed: []
    why: "Classification + action family (Reroute/Truncate/Stitch) matched observed transition dynamics"
  VersionBump:
    from: "v1.0.1"
    to: "v1.0.1"

5) Failure Mode Trace (required)

Trace:
Z4 legitimacy split → Drift velocity ↑ → Buffer ↓ → RateDominance breach (R≥1)
→ Narrative shifts from dynasty to sovereignty (NIT soft_lock) → TTC collapses
→ Controlled truncation of old sovereignty node → stitching into successor regime
→ treaty normalization acts as “stitch hardener” → P1→P2 stabilization

What we learn (portable, deployable rules)

1) Some collapses are not “defend & restore” problems — they’re reroute problems

If the failing organ is legitimacy/sovereignty, “more defense” doesn’t repair it.
The correct action family becomes: Reroute → Truncate → Stitch.

2) NIT is the hidden accelerator in regime transitions

Once narrative shifts to a new frame (sovereignty/assembly), reversal gets expensive fast.
CivOS implication: when NIT ≥ soft_lock, TTC estimation must compress even if physical fighting pauses.

3) Treaties function like stitching clamps

They reduce drift variance by converting de-facto control into de-jure stability. (That’s “stitch hardening.”)

Chhatrapati shivaji maharaj – Artofit

Backtest #3 — Chhatrapati Shivaji Maharaj night strike on Shaista Khan at Pune (Template C: Decision Quality / Actuation)

Frame: This is a pure FenceOS actuation backtest: under severe pressure from the Mughal Empire, can CivOS pick the correct action family (Truncate) and correctly predict what that actuation will change (TTC, R, legitimacy/morale), without pretending it “wins the war”?

Anchor facts (for context): Shaista Khan arrives in the Deccan (Jan 1660), takes Pune, and uses Shivaji’s Lal Mahal as his base; the night attack occurs 5 April 1663 with a ~400-man strike force; Shaista Khan is wounded and later transferred away (to Bengal, starting 1664). (Wikipedia)

Time window

  • t0: 1663-04-04 (one day before the strike)
  • t1: 1664-03-30 (Shaista Khan in Bengal office window anchor) (Wikipedia)

1) InputAt_t0 (allowed information)

CivOS.Backtest.C.DecisionQuality.v1:
  Meta:
    CaseID: "BT-SHI-1663-001"
    Domain: "conflict|city|asymmetric-actuation"
    Place: "PUNE/DECCAN"
  Time:
    t0: "1663-04-04"
    t1: "1664-03-30"
    Δt_days: 361
  InputAt_t0:
    SituationSummary: "Mughal pressure high; command node entrenched in Pune; open battle disadvantage; need TTC extension + morale/command disruption."
    AvailableLevers:
      Operator: ["small-unit action", "mobility", "local terrain/corridor knowledge", "rapid withdrawal"]
      Oracle:   ["intel on routines/access corridors", "risk estimation", "post-action signal tracking"]
      Visionary:["strategic intent: buy time, force enemy reallocation, preserve core forts/pipelines"]

2) RunState (pre-commit)

CivOS.RunState.v1:
  RunID: "RUN-BT-SHI-1663-001"
  Place: "PUNE/DECCAN"
  Window:
    t0: "1663-04-04"
    t1: "1664-03-30"
    Δt_days: 361
  Coordinates:
    Phase: "P1"
    Zoom: "Z3"
    Lane: "LANE.command-node-truncation"
    RoleFocus: "Mixed"
  Sensors:
    TTC_days: 45
    Buffer:
      thickness: 0.28
      trend: "down"
    Drift:
      velocity: 0.62
      acceleration: 0.12
    RateDominance:
      R: 1.15
      regime: "breach"
    NIT:
      value: 0.55
      regime: "soft_lock"
    TVIS:
      Hope: 0.60
      Grind: 0.55
      Wisdom: 0.60
      imbalance: 0.20
    Coupling:
      cross_lane_weight: 0.75
      contagion_risk: 0.70
  Decision:
    PrimaryAction: "Truncate"
    SecondaryActions: ["Fence", "Reroute"]
    TargetPhase: "P2"
    TargetZoom: "Z4"
    ExecutionRoles:
      Operator:
        - "OP.strike-command-node: create immediate command shock without prolonged engagement"
        - "OP.exit-fast: minimize losses; preserve pipeline continuity"
      Oracle:
        - "OR.access-corridor-map: identify low-friction entry/exit windows"
        - "OR.effect-monitor: track whether enemy reassigns command / tightens cost structure"
      Visionary:
        - "VI.TTC-extension-plan: use gained time to thicken buffer (forts, finance, coordination)"
        - "VI.counter-lock: prevent enemy narrative from hard-locking inevitability"
  Confidence:
    CoordConfidence: 0.76
    ActionConfidence: 0.70
    Notes: "When R breach + open-battle disadvantage, command-node truncation can raise TTC even if war continues."

Prediction commitment (what this RunState asserts)

By t1, expect:

  • A visible command shock (injury/embarrassment) that forces reallocation / removal of the entrenched commander, raising TTC for the defender.
  • Not “victory,” but TTC extension + morale/legitimacy disruption.

3) ObservedAt_t1 (what happened)

Within the window:

  • Night attack on 5 April 1663: ~400 attackers, Shaista Khan wounded (loss of fingers reported), his son killed; the force escapes. (Wikipedia)
  • After the humiliation, Aurangzeb transfers Shaista Khan away; his Bengal tenure begins in 1664. (Wikipedia)
ObservedAt_t1:
  WhatActuallyHappened:
    - "Successful command shock: target wounded; close associates killed; escape achieved"
    - "Strategic consequence: commander transferred away (Deccan → Bengal)"
  OutcomeNotes: "Truncation achieved: reduced command stability and imposed reallocation cost on attacker."

4) Verification (score + update)

CivOS.VerifyLog.v1:
  RunID: "RUN-BT-SHI-1663-001"
  ObservedOutcome:
    happened: "Night strike wounded commander + killed son + forced transfer away"
    outcome_class: "correct"
  Scores:
    CoordScore: 86
    ActionScore: 84
    TimingScore: 80
    TotalScore: 84
  FailureTag: "FT.NONE"
  ThresholdUpdate:
    changed: "yes"
    what_changed:
      - "Decision.PrimaryAction.v1: add Truncate pattern for 'R breach + open-battle disadvantage + accessible command node'"
      - "Classifier.Lane.v1: formalize LANE.command-node-truncation as a reusable lane"
      - "Verification metric: require 'TTC extension evidence' (enemy reallocation/transfer) post-truncate"
    why: "This is a reusable successful actuation pattern; codify it as a standard lane+decision trigger."
  VersionBump:
    from: "v1.0.1"
    to: "v1.0.2"

(Anchor facts cited for the observed outcomes.) (Wikipedia)

5) Failure mode trace (required)

Trace:
R breach + buffer thinning → open-battle infeasible → choose Truncate via command-node shock
→ attacker command stability drops → attacker pays reallocation cost (transfer/reassignment)
→ defender TTC increases (time bought) → defender can Stitch buffers (forts/finance/coordination)

What we learn (portable, “Conflict.OS deployable”)

  1. Truncate doesn’t mean “win.” It means stop the steepest damage slope long enough to restore regeneration.
  2. Command-node truncation is a valid lane when you can’t out-grind an attrition machine.
  3. Success metric is TTC extension evidence (reallocation, transfer, tightening costs), not body-count narratives.
  4. This pattern generalizes cleanly to office / politics / family as: “remove the unstable control node / break the coercive routine corridor,” without escalating into full collapse.

Backtest #4 — Decline slice of the Mughal Empire via Nader Shah’s invasion (Template B: Early Detection + Rate Dominance)

Frame: At t0, can CivOS flag that the system is already in RateDominance breach such that a high-tempo external attacker will force a near-term capital/treasury fracture (even if the empire “still exists” formally)?

Time window

  • t0: 1738-05-01 (invasion begins “in May 1738”) (Wikipedia)
  • t1: 1739-03-31 (Delhi is occupied/sacked in March 1739; Battle of Karnal 1739-02-24) (Wikipedia)
  • Δt: 334 days

1) InputAt_t0 (allowed information)

CivOS.Backtest.B.EarlyDetection.v1:
  Meta:
    CaseID: "BT-MUG-1738-001"
    Domain: "conflict|governance|capital-security"
    Place: "DELHI/NORTH-INDIA"
  Time:
    t0: "1738-05-01"
    t1: "1739-03-31"
    Δt_days: 334
  InputAt_t0:
    WeakSignals:
      - "A major external invasion has started; attacker tempo is high."
      - "Capital capture risk is now non-theoretical within < 1 year."
      - "If capital is taken, treasury + legitimacy shock becomes the dominant damage slope."
      - "Expect decisive-battle trap: a single defeat can collapse TTC rapidly."
    Noise:
      - "Imperial institutions still exist; formal sovereignty creates false buffer signals."
    AssumedPriors:
      coupling: 0.80
      buffer: 0.35

2) RunState (commit at t0)

CivOS.RunState.v1:
  RunID: "RUN-BT-MUG-1738-001"
  Place: "DELHI/NORTH-INDIA"
  Window:
    t0: "1738-05-01"
    t1: "1739-03-31"
    Δt_days: 334
  Coordinates:
    Phase: "P1"
    Zoom: "Z4"
    Lane: "LANE.capital-treasury-legitimacy"
    RoleFocus: "Mixed"
  Sensors:
    TTC_days: 210
    Buffer:
      thickness: 0.30
      trend: "down"
    Drift:
      velocity: 0.68
      acceleration: 0.14
    RateDominance:
      R: 1.12
      regime: "breach"
    NIT:
      value: 0.55
      regime: "soft_lock"
    TVIS:
      Hope: 0.45
      Grind: 0.55
      Wisdom: 0.40
      imbalance: 0.20
    Coupling:
      cross_lane_weight: 0.80
      contagion_risk: 0.78
  Decision:
    PrimaryAction: "Reroute"
    SecondaryActions: ["Fence", "Truncate"]
    TargetPhase: "P2"
    TargetZoom: "Z4"
    ExecutionRoles:
      Operator:
        - "OP.avoid-decisive-battle: do not accept a single-battle TTC collapse"
        - "OP.capital-fence: harden evacuation, continuity of command, and treasury protection"
        - "OP.loss-slope-control: reduce Ḋ by preventing concentrated shocks"
      Oracle:
        - "OR.TTC-refresh-weekly: update TTC using tempo + corridor loss signals"
        - "OR.detect-decisive-battle-trap: flag when one engagement can set R>>1"
        - "OR.buffer-audit: treat formal sovereignty as weak buffer if capital corridor is exposed"
      Visionary:
        - "VI.continuity-plan: preserve state-organ continuity even if capital falls"
        - "VI.rebuild-stitch: plan post-shock stitching to restore legitimacy and revenue pipelines"
  Confidence:
    CoordConfidence: 0.74
    ActionConfidence: 0.66
    Notes: "When invasion begins and coupling is high, the correct objective is TTC protection + loss-slope control, not prestige battle."

Prediction commitment (what this RunState asserts)

By t1, expect one of:

  • Decisive defeat producing rapid TTC collapse, followed by capital occupation/sack, or
  • Successful reroute/fence preventing capital fracture (the falsifier).

3) ObservedAt_t1 (what happened)

  • Battle of Karnal: 1739-02-24 (decisive defeat) (Wikipedia)
  • Delhi occupied/sacked: March 1739 (Wikipedia)
ObservedAt_t1:
  WhatActuallyHappened:
    - "Decisive defeat at Karnal (1739-02-24)"
    - "Capital occupied and sacked in March 1739"
  OutcomeNotes: "Exactly the decisive-battle TTC collapse + capital/treasury/legitimacy shock path."

4) Verification (score + update)

CivOS.VerifyLog.v1:
  RunID: "RUN-BT-MUG-1738-001"
  ObservedOutcome:
    happened: "Defeat at Karnal (1739-02-24) followed by Delhi occupation/sack (March 1739)"
    outcome_class: "correct"
  Scores:
    CoordScore: 86
    ActionScore: 72
    TimingScore: 78
    TotalScore: 79
  FailureTag: "FT.NONE"
  ThresholdUpdate:
    changed: "yes"
    what_changed:
      - "Decision.PrimaryAction.v1: if invasion_started AND Coupling>=0.75 -> prefer Reroute+Fence over prestige battle"
      - "Sensor.TTC.v1: add 'decisive-battle trap' compression (TTC := min(TTC, 120d) when single engagement can set R>>1)"
      - "Classifier.Lane.v1: lock LANE.capital-treasury-legitimacy as dominant lane when capital corridor exposed"
    why: "This is a reusable collapse pattern: decisive defeat → capital shock → legitimacy/treasury fracture."
  VersionBump:
    from: "v1.0.2"
    to: "v1.0.3"

5) Failure mode trace (required)

Trace:
Invasion begins (tempo high) → Coupling high → Buffer trend down → R enters breach
→ decisive-battle trap forms (single engagement can set R>>1) → TTC collapses
→ capital corridor fails → occupation/sack → legitimacy/treasury shock dominates damage slope

What we learn (portable, Conflict.OS deployable)

  1. The decisive-battle trap is a sensorable pattern. If one engagement can turn R from ~1.1 to “catastrophic,” TTC must be compressed immediately. (Wikipedia)
  2. When the capital/treasury lane becomes dominant, the objective changes: don’t “win glory,” preserve continuity and reduce loss slope. (Wikipedia)
  3. “Empire still exists” is not buffer. Formal sovereignty can mask real corridor exposure; buffer must be measured by controllable throughput.
Bajirao I birth anniversary: Facts about the Maratha warrior

Backtest #5 — Rise slice of the Maratha Empire under Bajirao I (Template A: Universal)

Frame: This is a rise-mode backtest: can CivOS classify a system moving P2 → P3 via buffer thickening + corridor reroute, and correctly predict what “success” looks like (not “final victory,” but measurable expansion + TTC extension + legitimacy/tax corridor capture)?

Time window

  • t0: 1720-04-17 (appointment as Peshwa by Shahu I) (Wikipedia)
  • t1: 1738-01-07 (Treaty of Bhopal signed) (Wikipedia)
  • Δt: 6474 days

1) InputAt_t0 (allowed information)

CivOS.Backtest.A.Universal.v1:
  Meta:
    CaseID: "BT-MAR-RISE-1720-001"
    Domain: "conflict|governance|expansion"
    Place: "DECCAN→NORTH-INDIA"
  Time:
    t0: "1720-04-17"
    t1: "1738-01-07"
    Δt_days: 6474
  InputAt_t0:
    EventStream:
      - timestamp: "1720-04-17"
        text: "New Peshwa appointed; strategic intent shifts toward offensive expansion against weakening imperial corridors."
        source: "chronicle/biography"
      - timestamp: null
        text: "Major threat actor in Deccan: Hyderabad/Nizam; conflict likely determines corridor control + tax rights."
        source: "political-military context"
    Constraints:
      known_unknowns:
        - "Exact timing of decisive engagements unknown at t0"
        - "External coalition responses unknown"

(Wikipedia)

2) RunState (commit at t0)

CivOS.RunState.v1:
  RunID: "RUN-BT-MAR-RISE-1720-001"
  Place: "DECCAN→NORTH-INDIA"
  Window:
    t0: "1720-04-17"
    t1: "1738-01-07"
    Δt_days: 6474
  Coordinates:
    Phase: "P2"
    Zoom: "Z4"
    Lane: "LANE.corridor-tax-sovereignty-expansion"
    RoleFocus: "Mixed"
  Sensors:
    TTC_days: 900
    Buffer:
      thickness: 0.55
      trend: "up"
    Drift:
      velocity: 0.30
      acceleration: 0.05
    RateDominance:
      R: 0.78
      regime: "safe"
    NIT:
      value: 0.42
      regime: "soft_lock"
    TVIS:
      Hope: 0.70
      Grind: 0.62
      Wisdom: 0.55
      imbalance: 0.18
    Coupling:
      cross_lane_weight: 0.75
      contagion_risk: 0.60
  Decision:
    PrimaryAction: "Reroute"
    SecondaryActions: ["Stitch", "Fence"]
    TargetPhase: "P3"
    TargetZoom: "Z4"
    ExecutionRoles:
      Operator:
        - "OP.mobility-corridor: win by tempo + encirclement rather than siege attrition"
        - "OP.tax-collection-throughput: convert battlefield wins into durable revenue corridors"
      Oracle:
        - "OR.track-corridor-capture: watch for Malwa/Gujarat/Delhi access signals"
        - "OR.detect-coalition-forming: if R rises toward warning, trigger Fence"
      Visionary:
        - "VI.institutionalize-peshwa-capital: turn operational mobility into stable governance throughput"
        - "VI.stitch-legitimacy: lock narrative from raids → durable political control"
  Confidence:
    CoordConfidence: 0.77
    ActionConfidence: 0.70
    Notes: "Rise-mode: keep R safe while turning tempo wins into permanent corridor + tax capture."

3) Prediction commitment (no hindsight)

PredictionCommitment:
  WhatWouldIExpectBy_t1:
    - "Decisive Deccan corridor win(s) against Hyderabad/Nizam forces"
    - "A northward demonstration that reaches the Mughal capital corridor (raid/pressure on Delhi)"
    - "A formal settlement that cedes/recognizes major territory (e.g., Malwa) and/or indemnity"
  WhatWouldFalsifyThis:
    - "Stalemate in Deccan (no corridor win) AND no northward reach by t1"
    - "R flips to breach early (collapse/fragmentation) instead of buffer thickening"

4) ObservedAt_t1 (what happened by the window end)

ObservedAt_t1:
  WhatActuallyHappened:
    - "Deccan corridor success against the Nizam (Palkhed campaign, Feb 1728, widely treated as decisive)"
    - "Northward reach: raid/pressure on Delhi in 1737"
    - "Formal settlement: Treaty of Bhopal signed 7 Jan 1738 (post–Dec 1737 battle), associated with Malwa concession and indemnity framing"
  OutcomeNotes: "Matches rise-mode pattern: corridor wins → capital reach → treaty ‘stitch clamp’."

(Wikipedia)

5) Verification (score + versioning)

CivOS.VerifyLog.v1:
  RunID: "RUN-BT-MAR-RISE-1720-001"
  ObservedOutcome:
    happened: "Deccan corridor win (Palkhed, 1728) → Delhi raid (1737) → Treaty of Bhopal (7 Jan 1738)"
    outcome_class: "correct"
  Scores:
    CoordScore: 86
    ActionScore: 78
    TimingScore: 72
    TotalScore: 80
  FailureTag: "FT.NONE"
  ThresholdUpdate:
    changed: "yes"
    what_changed:
      - "Add Rise-Mode rule: when R is safe and Buffer trend is up, prefer Reroute+Stitch (not Truncate)"
      - "Treaties-as-stitch-clamps: formal settlements reduce drift variance after corridor gains"
    why: "Successful positive pattern should be codified as reusable ‘rise loop’ rather than treated as anomaly."
  VersionBump:
    from: "v1.0.3"
    to: "v1.0.4"

(Encyclopedia Britannica)

6) Failure mode trace (required)

Trace (Rise):
Buffer trend up + R safe → choose Reroute (new corridors) + Stitch (institutionalize gains)
→ Deccan corridor win reduces threat pressure → northward reach increases leverage
→ treaty acts as stitch-clamp → stabilizes captured throughput → P2→P3 consolidation

What we learn from #5 (portable rules)

  1. Rise is measurable: “win” is not a vibe; it’s R safe + buffer up + corridor capture + treaty clamp. (Encyclopedia Britannica)
  2. Reroute is the rise action: expansion works when it converts tempo into durable tax/legitimacy corridors (stitch), not just raids. (Encyclopedia Britannica)
  3. Treaties are stitching hardeners: they reduce variance after gains (same logic you saw in the Ottoman backtest). (Wikipedia)

CivOS Meta-Learning Pack v1.0

(Canonical compression from Backtests #1–#5 → reusable patches + minimum set for any future news cycle)

This page converts the 5 backtests into portable rules that your OS can reuse anywhere (war, politics, office conflict, family conflict, city ops, education).

Backtest anchors (one-time references):

  • #1 Fall of Singapore (1942) — wrong zoom + “fortress narrative” bias
  • #2 Ottoman Empire transition — legitimacy reroute + treaty stitching
  • #3 Chhatrapati Shivaji Maharaj — command-node truncation actuation
  • #4 Mughal Empire decline slice — decisive-battle trap → capital/treasury fracture
  • #5 Maratha Empire rise slice — reroute + stitch as rise loop

1) What the 5 backtests teach (compressed invariants)

Invariant A — Wrong Zoom is the #1 catastrophic error

If the true driver is theatre control / logistics / legitimacy, but you model as local fortification / local incidents, your TTC and Buffer become fantasy.

Invariant B — NIT (narrative lock) compresses TTC

When meaning locks (“impregnable”, “cannot reverse”, “dynasty is over”, etc.) rise, the system becomes harder to unwind even if fighting pauses.

Invariant C — Decisive-battle trap is a measurable pattern

If one engagement can flip R from mild breach to catastrophic, TTC must be compressed immediately.

Invariant D — Truncate ≠ win; Truncate = slope control

Successful truncation buys time and optionality (TTC extension evidence), not “final victory”.

Invariant E — Treaties are stitch clamps

They “harden” a stitch by converting de-facto control into de-jure stability, reducing variance/drift.

Invariant F — Rise loop is: R safe + Buffer up + Corridor capture + Stitch

Rise is not vibes; it’s a measurable state transition.

2) RulePatch Library v1.0

(copy-paste; these are the reusable patches your OS now “knows”)

CivOS.MetaRulePatches.v1_0:
  # P1: Zoom Escalation Rule (prevents fortress-frame failure)
  - id: RP.ZOOM.001
    when:
      - Coupling.cross_lane_weight >= 0.60
      - "dominant driver is air/sea/logistics OR legitimacy OR capital corridor"
    set:
      Classifier.Zoom: "Z3_or_Z4"
    why: "wrong zoom inflates buffer + TTC"
  # P2: Lane Dominance Rule (choose the real load-path)
  - id: RP.LANE.001
    when:
      - "capital corridor exposed OR treasury/legitimacy shock plausible"
    set:
      Classifier.Lane: "LANE.capital-treasury-legitimacy"
    why: "this lane dominates damage slope"
  # P3: NIT Clamp Rule (narrative irreversibility compresses TTC)
  - id: RP.NIT.001
    when:
      - NIT.regime in ["soft_lock","hard_lock"]
    set:
      Sensor.TTC_days: "min(TTC_days, TTC_days*0.5)"
    why: "meaning-lock reduces reversibility"
  # P4: Fortress Narrative Bias Rule (detect & neutralize)
  - id: RP.NIT.002
    when:
      - "absolute claims of invulnerability / inevitability appear under load"
    set:
      Sensor.NIT.value: "max(NIT.value, 0.45)"
      Decision.PrimaryAction: "Fence_or_Truncate_review"
    why: "prevents buffer hallucination"
  # P5: Decisive-Battle Trap Compression
  - id: RP.TTC.001
    when:
      - "single engagement can set R >> 1"
    set:
      Sensor.TTC_days: "min(TTC_days, 120)"
      Decision.PrimaryAction: "Fence_or_Reroute"
    why: "avoid one-hit TTC collapse"
  # P6: Command-Node Truncation Pattern (asymmetric actuation)
  - id: RP.ACTION.001
    when:
      - RateDominance.regime == "breach"
      - "open-battle disadvantage"
      - "accessible command node exists"
    set:
      Classifier.Lane: "LANE.command-node-truncation"
      Decision.PrimaryAction: "Truncate"
      VerifyMetric: "require TTC extension evidence (enemy reallocation/transfer)"
    why: "slope control without prolonged engagement"
  # P7: Legitimacy Transition Pattern (restoration is wrong objective)
  - id: RP.ACTION.002
    when:
      - Classifier.Lane == "LANE.legitimacy-sovereignty-transition"
      - RateDominance.regime in ["warning","breach"]
    set:
      Decision.PrimaryAction: "Reroute"
      SecondaryActions: ["Truncate","Stitch"]
    why: "successor structure beats restoration"
  # P8: Treaty-as-Stitch Clamp
  - id: RP.STITCH.001
    when:
      - "formal settlement / treaty / legal normalization occurs"
    set:
      Drift.acceleration: "down"
      Buffer.trend: "up"
    why: "stitch hardening reduces variance"
  # P9: Rise Loop Trigger
  - id: RP.RISE.001
    when:
      - RateDominance.regime == "safe"
      - Buffer.trend == "up"
    set:
      Decision.PrimaryAction: "Reroute"
      SecondaryActions: ["Stitch","Fence"]
      TargetPhase: "P3"
    why: "convert tempo into durable corridors"
  # P10: Closure Enforcement (the OS must learn)
  - id: RP.CLOSE.001
    when:
      - VerifyLog.TotalScore < 70
    require:
      - "FailureTag assigned"
      - "ThresholdUpdate.changed = yes"
      - "VersionBump forward"
    why: "prevents non-learning analysis"

3) Failure Tags that matter most (priority order)

This is your “most expensive mistakes” list.

  1. FT.WRONG_ZOOM — produces the worst TTC/Buffer hallucinations
  2. FT.TTC_MISREAD — you can be right but useless (late)
  3. FT.WRONG_LANE — correct instincts, wrong load-path = wrong action
  4. FT.NIT_MISREAD — you miss irreversibility and overestimate reversibility
  5. FT.HIDDEN_COUPLING — contagion spreads and you think it’s local
  6. FT.WRONG_ACTION — you detected the regime, but chose wrong actuation family

4) Minimum Backtest Set for any future conflict/news cycle

If you only run 5 backtests to calibrate a new domain, use these archetypes:

  1. Fortress-frame failure (tests WRONG_ZOOM)
  2. Legitimacy transition (tests REROUTE vs RESTORE)
  3. Asymmetric truncation (tests actuation success metric = TTC extension evidence)
  4. Decisive-battle trap / capital corridor (tests TTC compression + lane dominance)
  5. Rise loop (tests positive-mode routing: R safe + Buffer up + Stitch)

Your current #1–#5 already cover this minimum set. That’s why your system is now genuinely “deployable”.

5) Deployable “Daily News / Conflict Loop” (10-minute operator run)

This is how you run CivOS every day without bloating:

Step 1 — Choose t0 (today) and Δt (7/30/90 days).
Step 2 — Produce RunState (Phase/Zoom/Lane + TTC/Buffer/R/NIT/TVIS).
Step 3 — Commit 2–3 falsifiable predictions (no hindsight).
Step 4 — Choose PrimaryAction + role tasks (Operator/Oracle/Visionary).
Step 5 — Schedule VerifyLog at t1 (or update weekly).
Step 6 — If score < 70 → patch + version bump.

Minimal output each day:

DailyRun.MinOutput.v1:
  Phase: <P0..P3>
  Zoom: <Z0..Z6>
  Lane: <LaneID>
  TTC_days: <number>
  R_regime: <safe|warning|breach>
  NIT_regime: <reversible|soft_lock|hard_lock>
  PrimaryAction: <Monitor|Fence|Truncate|Stitch|Reroute|Escalate>
  3_Predictions: [..]
  2_Falsifiers: [..]

Start Here:

Start here if you want the full sequence:

Vocabulary OS Series Index:
https://edukatesg.com/vocabulary-os-series-index/

Fence English Learning System: 

eduKateSG Learning Systems: 

Recommended Internal Links (Spine)

Start Here for Lattice Infrastructure Connectors