Skip to content

analytical-supplementary-methodology.md

Latest commit

 

History

History
383 lines (291 loc) · 25.1 KB

analytical-supplementary-methodology.md

File metadata and controls

383 lines (291 loc) · 25.1 KB

Analytical Supplementary Methodology

Version: v2.0 · Effective date: 2026-04-25 · Owner: Intelligence Operative role · Review cycle: Quarterly

Scope. Governs the Analytical Supplementary family of optional deep-dive templates that can be produced alongside the 23 mandatory artifacts. These templates are never required by 05-analysis-gate.md and never replace a core artifact — they augment the core set with specialised analytic lenses when a run's scope warrants it.

Sister documents:

🎯 AI-FIRST Methodology Card

🚦 Read this card before writing a single paragraph. It names the artifact this methodology owns, the gate check it satisfies, the evidence-density target it must hit, and the Pass-1 / Pass-2 discipline required by .github/copilot-instructions.md §5 (AI-FIRST Quality Principle).

Field Value
Purpose Family-S supplementary methodologies — PESTLE, political-STRIDE, wildcards/black-swans, quantitative SWOT — produced for long-horizon (year / cycle) and high-impact runs.
Inputs Family A synthesis; Family C strategic extensions; long-horizon scenario sets
Outputs pestle-analysis.md, political-stride-assessment.md, wildcards-blackswans.md, quantitative-swot.md
Owning artifact(s) Family-S supplementary (AS1..AS4)
Owning gate check Check 11 (supplementary) — non-blocking for short horizons; blocking for year/cycle per reference-quality-thresholds.json#analyticalSupplementary
Citation density target Per-dimension table: ≥ 1 evidence anchor per row (PESTLE 4 dimensions, STRIDE 6 categories, wildcards ≥ 8 entries, qSWOT ≥ 3 per quadrant)
Banned phrases Enforced via political-style-guide.md §Machine-readable banned-phrase list
Threshold source reference-quality-thresholds.json → thresholds[articleType][artifact] (fallback defaults.coreArtifactFloor)

✅ Pass-1 checklist (creation — minimal viable artifact)

  • PESTLE: ≥ 4 dimension tables and ≥ 3 cross-dimension interactions
  • STRIDE: ≥ 3 entries per dimension, ≥ 2 attack trees, ≥ 5 TTP rows
  • Wildcards/black-swans: ≥ 8 wildcards + ≥ 3 black-swans + ≥ 2 consequence trees
  • Quantitative SWOT: ≥ 3 entries per quadrant + ≥ 3 sensitivity runs + ≥ 1 TOWS action per quadrant
  • Produce every required sub-section listed in the owning template
  • Add ≥ 1 evidence anchor (dok_id, vote id, named MP, or primary-source URL) per analytical claim
  • Apply the correct WEP confidence band for the run's horizon (72h / week / month / quarter / year / cycle)
  • Include ≥ 1 themed Mermaid diagram with style … or themeVariables config (where structurally meaningful)
  • Cross-link the relevant template under analysis/templates/ and the gate check it satisfies

🔁 Pass-2 checklist (read-back & improve — AI-FIRST mandatory)

  • Verify each supplementary artifact's row floor in reference-quality-thresholds.json#analyticalSupplementary.*.rowFloor
  • For year/cycle: confirm wildcards include ≥ 1 climate / ≥ 1 geopolitical / ≥ 1 technological / ≥ 1 economic / ≥ 1 social tag
  • Re-read the file end-to-end; flag every claim that lacks an evidence anchor and add one
  • Replace every banned phrase listed in political-style-guide.md §Machine-readable banned-phrase list with an evidence-anchored alternative
  • Tighten WEP language: never above likely without ≥ 3 cycle-aged sources for year/cycle horizons
  • Strengthen Mermaid (color-coded style … directives, themeVariables, ≥ 5 nodes where the structure admits it)
  • Add ≥ 1 second-order effect, cui-bono note, or counterfactual where the artifact admits one
  • Verify citation density meets the per-file target below and the gate's evidence-density rules

🟢 Exemplar (good — pattern-match this)

(STRIDE row) "Tampering — Legislative integrity: actor=foreign IO; vector=disinformation amplification; evidence=H902KU14 2026-04-19 motion citing X-platform integrity logs ([B2]); mitigation=Statskontoret oversight."

🔴 Anti-exemplar (failure mode — never ship this)

(failure mode) "Wildcards include various scenarios that might affect Swedish politics." — < 8 entries, no tags, no evidence, banned phrases.

🔗 Cross-links

🎯 Purpose

Provide Riksdagsmonitor analysts with rigorously defined optional deep-dive templates that map 1-to-1 to analytical frameworks explicitly listed in the intelligence-operative agent's "Core Expertise" and "Analytical Frameworks" sections but that were not previously templated:

  • PESTLE — macro-environment scan (political, economic, social, technological, legal, environmental)
  • STRIDE-political — adapted STRIDE threat model for political/electoral/institutional surfaces
  • Wildcards & Black-Swans — low-probability / high-impact tail complementing scenario-analysis.md
  • Quantitative SWOT — numerical DIW × confidence × leverage scoring on top of narrative SWOT

📦 Outputs (4 templates)

ID Template Pairs with Triggered when
AS-1 pestle-analysis.md swot-analysis.md, risk-assessment.md, scenario-analysis.md event crosses ≥ 2 PESTLE dimensions
AS-2 political-stride-assessment.md threat-analysis.md, risk-assessment.md (Institutional/Corruption) election-adjacent, integrity incident, disinfo spike, critical-infra vote
AS-3 wildcards-blackswans.md scenario-analysis.md, forward-indicators.md long-horizon forecasting (monthly-review, election-year aggregation)
AS-4 quantitative-swot.md swot-analysis.md, significance-scoring.md, executive-brief.md decision memo requiring scored ranking

🗂️ Decision tree — when to produce each template

Use this decision tree at the start of every run to determine which (if any) Analytical Supplementary templates to produce. Producing templates unnecessarily bloats runs and dilutes signal.

1. Is this a standard daily run with ≤ 3 documents and no election-adjacent or cross-domain flag?
   YES → skip all AS templates; the 23 core artifacts are sufficient
   NO  → continue to step 2

2. Does the trigger event affect ≥ 2 PESTLE dimensions (P/E/S/T/L/Env)?
   YES → produce PESTLE (AS-1)
   NO  → skip AS-1

3. Is the trigger election-adjacent (within 180 days of val-dag), OR does it involve an
   integrity incident / disinformation spike / critical-infra vote?
   YES → produce STRIDE-political (AS-2)
   NO  → skip AS-2

4. Is the assessment horizon ≥ 6 months AND does the topic involve cascading risk or
   structural uncertainty beyond the 3–5 main scenarios?
   YES → produce Wildcards & Black-Swans (AS-3)
   NO  → skip AS-3

5. Does the consumer of this run need a numerically ranked recommendation
   (coalition negotiation, budget allocation, party strategy, election forecast)?
   YES → produce Quantitative SWOT (AS-4); ALWAYS produce swot-analysis.md first
   NO  → skip AS-4

Priority rule: if all four apply (rare), prioritise:
  - Security/integrity topics → AS-2 + AS-3
  - Policy analysis topics   → AS-1 + AS-4

🧭 When NOT to produce

  • Do not use an Analytical Supplementary template as a substitute for a missing core artifact. The gate will still fail.
  • Do not proliferate templates for a "standard" daily morning run with ≤ 3 documents — the core 23 are enough.
  • Do not produce AS-4 (Quantitative SWOT) without first completing swot-analysis.md — the QSWOT scores are meaningless without the narrative backbone.
  • Do not produce AS-3 (Wildcards) for a single-document interpellations run — the tail analysis requires a minimum evidence base of ≥ 5 primary sources.
  • Do not produce AS-2 (STRIDE-political) for purely economic topics where institutional integrity is not at stake.

🪢 Composition rules

Rule 1 — Single DIW weight vector

quantitative-swot.md MUST use the same weight vector as significance-scoring.md:

w_D = 0.35  (Decision relevance)
w_I = 0.25  (Information novelty)
w_W = 0.20  (Wave / momentum)
w_S = 0.20  (Stakeholder reach)

Any deviation from these weights requires updating both files simultaneously and noting the justification in methodology-reflection.md. The gate check does not yet auto-enforce this, but reviewers must flag mismatches during Pass-2 audit.

Rule 2 — Every cell cites primary evidence

PESTLE / STRIDE / Wildcard / QSWOT cells that carry an assertion MUST cite a dok_id, primary URL host (riksdagen.se, regeringen.se, scb.se, data.imf.org, worldbank.org, eur-lex.europa.eu, known myndighets domain), or named source with date. The same rule applies as in swot-analysis.md (gate check 4). Assertions without evidence citations are classified as [UNSUPPORTED] and must be remediated in Pass-2 or removed.

Rule 3 — PESTLE / Wildcards feed Forward-Indicators

Every PESTLE indicator row or Wildcard trigger indicator MUST surface in at least one horizon section of forward-indicators.md with a dated trigger (contributes to check 8 floor of ≥ 10 dated indicators). Analyst must explicitly copy the indicator row and link it.

Rule 4 — STRIDE-political maps to threat-analysis.md TTPs

Every STRIDE row with Likelihood × Impact ≥ 12 MUST appear in the canonical MITRE-style TTP mapping of threat-analysis.md. STRIDE adds adversary-model structure; threat-analysis.md remains the canonical kill-chain file. Rows below the 12-threshold are optional cross-references.

Rule 5 — Quantitative SWOT never replaces narrative SWOT

swot-analysis.md remains the Pass-2-enforced narrative artifact. quantitative-swot.md is read-alongside. Top-3 scored items (highest Strength, highest Opportunity, most-negative Threat) MUST surface in executive-brief.md § 3 Decisions with item ID citation.

Rule 6 — Wildcards cite historical analogues

Every wildcard (W1–Wn) must cite at least one historical analogue (Swedish or comparable democracy) in the wildcards-blackswans.md §Historical analogues table. Black-swan candidates (BS1–BSn) must cite a "why under-weighted" cognitive/structural bias explanation — pure rarity is not sufficient justification.

Rule 7 — IMF-first for all economic evidence

All economic data in any AS template (PESTLE E dimension, QSWOT S/W/O/T economic items, Wildcard W2/W8 fiscal triggers, STRIDE D fiscal impact) cites IMF as the source — see imf-indicator-mapping.md and .github/aw/ECONOMIC_DATA_CONTRACT.md v3.0. SCB supplies Swedish-specific ground truth where the AS template needs high-frequency or regional detail.

🔬 Per-template analytic moves

PESTLE

Inputs required before starting:

  • session-baseline.md for current legislative calendar
  • swot-analysis.md §External for the external environment baseline
  • IMF WEO vintage within 6 months of article date
  • SCB table codes identified for Social / Environmental dimensions
  • ≥ 1 named myndighet report per relevant dimension

Analytic moves (in order):

  1. Declare scope (trigger event, horizon, unit of analysis, source set) — failure to scope produces unfocused output
  2. Fill 6 dimension tables: P → E → S → T → L → E_env; minimum 5 rows per dimension for Comprehensive tier
  3. For each Economic row, cite IMF WEO indicator and vintage in format WEO:INDICATOR_CODE (WEO-YYYY-MM)
  4. For each Social/Environmental row, cite SCB table code or World Bank retained indicator
  5. Write 1–3 sentence key judgement per dimension with WEP tag — do NOT copy row content verbatim
  6. Write dissent note per dimension naming a competing interpretation
  7. Populate cross-dimension interactions table (≥ 3; ≥ 5 for Comprehensive); rank by magnitude
  8. Produce PESTLE heatmap mermaid chart with numeric risk levels
  9. Complete Election 2026 PESTLE lens if within 180 days of val-dag OR if article type is election-2026-analysis
  10. Populate PIR feedback

Anti-patterns (cause Pass-2 failure):

  • ❌ Citing "general context" or "well-known background" without a specific source
  • ❌ Using World Bank GDP/unemployment/inflation codes instead of IMF equivalents
  • ❌ Key judgement that merely restates row content without synthesis
  • ❌ Missing WEP tag on any probabilistic claim
  • ❌ Cross-dimension interactions using vague "affects" language — must specify mechanism, direction, and magnitude
  • ❌ Producing PESTLE for a single-dimension issue (use risk-assessment.md §External instead)
  • ❌ Electoral lens omitted when article is within 180 days of val-dag

STRIDE-political

Inputs required before starting:

  • threat-analysis.md (to align MITRE TTP labelling)
  • risk-assessment.md §Institutional (to avoid duplicating content; STRIDE should add adversary-model depth, not copy rows)
  • Named adversary model with at least: motivation, capability level 1–5, known or assessed technique
  • Scoped entity and trust boundary clearly defined

Analytic moves (in order):

  1. Declare scope: entity, trust boundary, time horizon, adversary model — specific; not generic
  2. Fill 6 STRIDE tables (S/T/R/I/D/E); minimum 4 rows per dimension; score L (1–5) × I (1–5) for every row
  3. Flag all rows with L × I ≥ 12 as mandatory cross-references to threat-analysis.md
  4. Produce ≥ 2 Mermaid attack trees, each with ≥ 3 node levels (Goal → tactic → technique → consequence); colour-code per standard
  5. Produce MITRE ATT&CK–style TTP table; mark 🚨 (mandatory cross-ref) or ⚠️ (advisory) for each row
  6. Produce recommended controls table mapped to ISO 27001:2022 + NIST CSF 2.0 + CIS Controls v8.1
  7. Produce Election 2026 STRIDE lens if within 180 days of val-dag
  8. Complete PIR feedback

Scoring rubric (L × I priority matrix):

L\I 1 (negligible) 2 (minor) 3 (moderate) 4 (significant) 5 (severe)
5 (likely+) 5 10 15 🚨 20 🚨 25 🚨
4 (about even) 4 8 12 ⚠️ 16 🚨 20 🚨
3 (unlikely) 3 6 9 12 ⚠️ 15 🚨
2 (very unlikely) 2 4 6 8 10
1 (remote) 1 2 3 4 5

Anti-patterns (cause Pass-2 failure):

  • ❌ Generic STRIDE table rows without named vectors and specific targets
  • ❌ Attack tree with fewer than 3 node levels or missing colour definitions
  • ❌ MITRE TTP table without priority classification (🚨 / ⚠️ / optional)
  • ❌ Controls table missing ISO 27001 + NIST CSF columns
  • ❌ Adversary model stated as "hostile actor" without capability / motivation specification
  • ❌ L×I ≥ 12 rows not cross-referenced to threat-analysis.md
  • ❌ Producing STRIDE for a purely fiscal topic where no institutional integrity dimension exists

Wildcards & Black-Swans

Inputs required before starting:

  • scenario-analysis.md (to confirm which scenarios are already covered in the nominal distribution)
  • forward-indicators.md (to check which early-warning indicators are already tracked)
  • risk-assessment.md §External (to avoid duplicating high-impact/high-probability rows; wildcards live in the tail)
  • historical-parallels.md (to source analogues)
  • Reference class estimate: what base-rate applies to "political wildcard" in a stable Nordic democracy?

Analytic moves (in order):

  1. Declare scope: horizon, domain filter, source set, reference class baseline, trigger
  2. Define ICD-203 terms (wildcard / black-swan / cascade / resilience) in file header
  3. Populate wildcard register (≥ 8 events): assign WEP, trigger indicator, monitoring source, lead time, impact vectors, counter-measures, Admiralty grade
  4. Populate black-swan candidates (≥ 3): explain "why under-weighted" (cognitive/structural bias), provide ≤ 4-step causal chain, name early-warning signal, estimate recovery lead time
  5. Produce ≥ 2 Mermaid cascading consequence trees (≥ 3 consequence layers each, colour-coded)
  6. Populate historical analogues table (≥ 1 per major wildcard)
  7. Populate early-warning indicators table; link each to forward-indicators.md horizon section
  8. Produce resilience assessment (5 dimensions): rate 1–5, cite evidence, identify gap, recommend improvement, assign priority
  9. Complete Election 2026 implications table if within 180 days of val-dag OR if monthly-review
  10. Complete PIR feedback

Anti-patterns (cause Pass-2 failure):

  • ❌ Wildcard events that are actually high-probability (> 45 %); those belong in scenario-analysis.md
  • ❌ Black-swan candidates without a "why under-weighted" cognitive/structural bias explanation
  • ❌ Causal chains with > 4 steps (becomes speculative; truncate or split into two candidates)
  • ❌ Cascade trees with fewer than 3 levels or no mitigation lever node
  • ❌ Wildcard trigger indicators not appearing in forward-indicators.md — Rule 3 violation
  • ❌ Resilience dimension scored without evidence citation (a score of "3" needs a named source)
  • ❌ Historical analogues absent from the template (prevents grounding of prior estimates)
  • ❌ All wildcards in the same domain (Political only); domain diversity rule ≥ 5 different domains required

Quantitative SWOT

Inputs required before starting:

  • swot-analysis.md fully completed (narrative backbone; QSWOT scores are undefined without it)
  • significance-scoring.md weight vector confirmed (w_D=0.35, w_I=0.25, w_W=0.20, w_S=0.20)
  • IMF WEO vintage for any economic item; SCB table codes for social/demographic items
  • Consumer of the analysis identified: coalition negotiators, party strategists, press office, external?

Analytic moves (in order):

  1. Declare scope, entity, perspective, assessment horizon, anchor date
  2. Document weight vector and confirm match with significance-scoring.md
  3. Document scoring rubric with WEP ↔ C mapping table
  4. Provide worked example (compute one S item end-to-end to validate formula)
  5. Score ≥ 3 Strengths; compute total; identify top item
  6. Score ≥ 3 Weaknesses; compute total; identify top item
  7. Score ≥ 3 Opportunities; compute total; identify top item
  8. Score ≥ 3 Threats; compute total; identify top item
  9. Compute composite SWOT position metrics (net position, SW balance, OT balance, high-confidence share)
  10. Write overall position narrative (2–3 sentences; cite specific scores; do NOT use vague "mixed picture" language)
  11. Populate TOWS 2×2 with ≥ 1 specific action per quadrant, each citing item IDs
  12. Run sensitivity analysis: 3–5 parameter flips; flag "fragile consensus" if net-position sign reverses
  13. Produce mermaid xychart-beta with actual computed scores (not sample values)
  14. Complete Election 2026 implications table if within 180 days of val-dag
  15. Complete PIR feedback; surface top-3 items to executive-brief.md

Formula validation checklist:

For every item row, verify:
  □ I ∈ [-5.0, +5.0]
  □ C ∈ [0.20, 0.95]
  □ L ∈ [0.10, 1.00]
  □ T ∈ [0.30, 1.00]
  □ dRel + iNov + wMom + sReach ≠ 1.0 (these are per-dimension sub-scores, not a probability sum)
  □ w_D·dRel + w_I·iNov + w_W·wMom + w_S·sReach ≤ 1.0
  □ Score sign matches item class (positive for S/O; negative for W/T)

Anti-patterns (cause Pass-2 failure):

  • ❌ Producing QSWOT without a completed swot-analysis.md companion
  • ❌ Using different weight vector from significance-scoring.md
  • ❌ Scores populated with sample values from the template instead of actual computed values
  • ❌ All items clustered in a narrow score range (2.0–2.5) — indicates parameter anchoring; vary parameters
  • ❌ TOWS cell containing generic text ("leverage strengths") without citing specific item IDs
  • ❌ Sensitivity analysis using only 1 parameter flip — minimum is 3
  • ❌ Net-position narrative that uses "mixed" or "nuanced" without specifying which specific items drive the sign
  • ❌ Mermaid chart showing sample bar values instead of the actual computed scores

📏 Depth floors (line counts)

Template Standard Deep Comprehensive / Tier-C
pestle-analysis.md 100 150 220
political-stride-assessment.md 110 160 240
wildcards-blackswans.md 110 160 240
quantitative-swot.md 110 160 240

Floors apply only when the template is produced. Not producing them is never a failure.

What "floor" means: the minimum number of content-bearing lines (excluding blank lines and the Pass-2 checklist). A file at exactly the floor has used the minimum scaffolding; Comprehensive runs are expected to exceed the floor by 30–50 %.

🛡 Tradecraft compliance

All four templates inherit osint-tradecraft-standards.md:

  • Admiralty A–F × 1–6 source-reliability grading in evidence columns (e.g. A1 = completely reliable / confirmed; B2 = usually reliable / probably true; C3 = fairly reliable / possibly true; D4 = not usually reliable / doubtful; E5 = unreliable / improbable; F6 = cannot be judged / truth cannot be judged)
  • WEP bands for every estimative claim — Remote / Very unlikely / Unlikely / About even / Likely / Very likely / Almost certain — per political-style-guide.md §WEP
  • ICD 203 standards of analytic tradecraft:
    1. Objectivity — present all relevant information regardless of its policy implications
    2. Independence — distinguish analytic judgements from policy preferences
    3. Timeliness — deliver analysis while it can still inform decisions
    4. Sources cited — attribute all information to primary sources
    5. Uncertainties conveyed — label gaps, assumptions, and alternative interpretations
    6. Distinguish underlying vs. analytic — separate observed facts from analytical inferences
    7. Relevance — relate every claim to the PIRs or key questions
    8. Logical argumentation — claims must follow from evidence with traceable reasoning
    9. Consistency — do not contradict other artifacts in the same run without explanation
    10. Accurate judgements of change — label when the assessment changes from prior runs
  • SAT catalog linkage — PESTLE is a SAT itself; STRIDE-political draws on Red-Team and Threat-Trees; Wildcards uses Foresight + Alternative-Futures; QSWOT uses Weighted-Ranking + Sensitivity Analysis

🔗 Cross-links