admiralty-rubric.md

May 15, 2026 · View on GitHub

Hack23 Logo

⚓ Admiralty Grading Rubric — Riksdagsmonitor

🎯 Worked-Example Sheet for Source Reliability (A–F) and Information Credibility (1–6)
🏛️ ICD 206 · NATO STANAG 2511 · Heuer & Pherson SAT (3rd ed.) · ≥80 % Inter-Analyst Agreement

Owner Version Effective Date Classification

📋 Document Owner: CEO | 📄 Version: 1.0 | 📅 Last Updated: 2026-05-14 (UTC) 🔄 Review Cycle: Quarterly | ⏰ Next Review: 2026-08-14 🏢 Owner: Hack23 AB (Org.nr 559534-7807) | 🏷️ Classification: Public

Purpose: Provide a worked-example sheet so that two independent Riksdagsmonitor analysts reach the same Admiralty grade ≥ 80 % of the time. Eliminates the B-vs-C and 2-vs-3 subjectivity noted in the 2026-05-14 methodology-reflection cohort. Cite this rubric whenever applying Admiralty grades in any analysis artifact. Wire: 05-analysis-gate.md Check 7 for methodology-reflection ICD 203 audit.

Authority: ICD 206 (Source Reliability and Credibility Standards), NATO STANAG 2511 (Intelligence Grading), Heuer & Pherson Structured Analytic Techniques for Intelligence Analysis (3rd ed.) ch. 6.


📐 The Admiralty 2×6 Grid

Every source citation in a Riksdagsmonitor analysis artifact must carry two components:

  1. Reliability grade (letter A–F) — assesses the source, not the information.
  2. Credibility grade (number 1–6) — assesses the information item, not the source.

Combined notation: [B2], [A1], [C3], etc.

🔠 Source Reliability (A–F)

GradeLabelDefinitionSwedish political-intelligence examples
ACompletely reliableSource has a long, documented track record; all prior information later confirmed; no known bias or agendaOfficial Riksdag chamber records (riksdagen.se); Regeringskansliet official publications; SCB open statistics; IMF WEO official dataset
BUsually reliableSource has been reliable in most cases; minor errors or delays; no systematic bias identifiedSwedish news agencies (TT, Reuters Sweden); peer-reviewed Swedish political science journals; official Lagrådet yttrande archive
CFairly reliableSource has provided reliable information often enough to be worth using, but has made significant errors or shown occasional biasSwedish broadsheets (DN, SvD, GP) with byline; identified academic researchers; NGO policy reports with named methodology
DNot usually reliableSource has provided unreliable or inaccurate information more often than not, but not entirely discountedPartisan think-tanks without transparent methodology; unnamed "political sources" in tabloid press; unverified social media accounts with history of inaccuracy
EUnreliableSource has consistently provided false, misleading, or unverifiable informationAnonymous internet commentary; known disinformation outlets; unverified telegram channels
FReliability cannot be judgedNewly encountered source; no track record; unable to assess independence or biasFirst-time whistleblower contact; unknown NGO; unidentified document

🔢 Information Credibility (1–6)

GradeLabelDefinitionSwedish political-intelligence examples
1Confirmed by other sourcesInformation independently corroborated by ≥2 unrelated sourcesVote outcome confirmed in both search_voteringar (Riksdag API) AND separate TT wire report
2Probably trueSource is known reliable AND content consistent with established patterns; no contradictory evidenceGovernment budget framework figure matches Finansdepartementet press release + IMF WEO prior-year
3Possibly truePlausible content from a fairly reliable source; some corroboration but not definitiveSingle broadsheet report of coalition-talks; no second source yet
4DoubtfulInformation conflicts with known facts or established patterns, but not impossibleReport claims SD is considering formal coalition entry — conflicts with SD's stated position but not ruled out
5ImprobableInformation is highly inconsistent with established facts or the source's known recordClaim that Statsminister privately supports EU membership suspension — no structural support
6Cannot be judgedInsufficient context to assess credibility; neither confirming nor denying evidence availableUnverified leak about internal ministry briefing on sensitive treaty — content plausible but unconfirmable

🔑 Decision trees for common Riksdagsmonitor source types

Tree 1 — Riksdag official records

Is the source riksdagen.se (API or web) for votes, dok_ids, or transcripts?
  YES → Reliability = A (completely reliable official record)
        Credibility = 1 (confirmed by primary record)
        Grade: [A1]
  NO  → Continue to Tree 2

Tree 2 — Government official publications

Is the source regeringen.se, Finansdepartementet, SCB, IMF WEO, or Lagrådet official site?
  YES → Reliability = A
        Is the information also cited in at least one independent secondary source?
          YES → Credibility = 1; Grade: [A1]
          NO  → Credibility = 2; Grade: [A2]
  NO  → Continue to Tree 3

Tree 3 — Swedish news agencies and established media

Is the source TT (Tidningarnas Telegrambyrå) or Reuters Sweden with byline?
  YES → Reliability = B
        Is the information corroborated by ≥1 independent source?
          YES → Credibility = 1 or 2; Grade: [B1] or [B2]
          NO  → Credibility = 3; Grade: [B3]
  Is the source DN, SvD, GP, Aftonbladet, Expressen, SVT, or SR with byline?
  YES → Reliability = B (established editorial standards)
        Is the information corroborated?
          YES → Credibility = 2; Grade: [B2]
          NO  → Credibility = 3; Grade: [B3]
  NO  → Continue to Tree 4

Tree 4 — NGOs, think-tanks, academia

Is the source a peer-reviewed academic publication or identified researcher?
  YES → Reliability = B
        Is the content methodologically sound and independently cited?
          YES → Credibility = 2; Grade: [B2]
          NO  → Credibility = 3; Grade: [B3]
Is the source a named NGO with transparent methodology?
  YES → Reliability = C
        Is the claim corroborated by independent source?
          YES → Credibility = 2; Grade: [C2]
          NO  → Credibility = 3; Grade: [C3]
Is the source a partisan think-tank or advocacy organisation?
  YES → Reliability = D
        Use only if no better source; always label credibility ≥ 4; Grade: [D4] minimum

📝 Worked examples (≥20)

Example 1 — Riksdag vote record

FieldValue
Sourcesearch_voteringar API — SD vote on FiU48 2026-04-24
Grade[A1]
Reliability rationaleOfficial Riksdag chamber record; no transcription or interpretation involved
Credibility rationalePrimary vote record confirmed; no ambiguity possible in digital vote ledger
Wrong grade[B2] — ❌ This would imply "usually reliable" for a primary digital record that is either correct or doesn't exist
Disambiguation noteIf the analyst is interpreting the vote pattern (e.g. "SD defection"), that interpretation receives its own grade ([B3] at best) separate from the vote fact ([A1])

Example 2 — IMF WEO Sweden NGDP_RPCH forecast

FieldValue
SourceIMF WEO April 2026 vintage — Sweden GDP growth forecast
Grade[A2]
Reliability rationaleIMF official flagship publication; no history of Swedish data errors; no known bias for Sweden
Credibility rationaleForecast figure is the IMF's own model output — not confirmed by an independent second source (IMF is the only source for its own projections)
Wrong grade[A1] — ❌ Credibility 1 requires independent corroboration; IMF forecast cannot be "confirmed" by another source that also uses IMF data
Disambiguation noteWhen SCB quarterly national accounts confirm IMF forecast within ±0.3 pp, upgrade to [A1]

Example 3 — DN article with named journalist

FieldValue
SourceDagens Nyheter (Ida Söderberg, 2026-05-12) — report on coalition talks
Grade[B3]
Reliability rationaleEstablished newspaper with editorial standards; named journalist with track record
Credibility rationaleSingle-source report; coalition talks claimed but not confirmed by Riksdag records or second outlet
Wrong grade[C3] — ❌ DN has long-established editorial standards that place it in reliability tier B, not C
Disambiguation noteIf TT also runs the story without attribution to DN, upgrade to [B2]

Example 4 — NGO healthcare access report

FieldValue
SourceVårdförbundet annual report on nurse staffing shortfall (2025)
Grade[C2]
Reliability rationaleNamed NGO with transparent survey methodology; professional body with institutional interest in the topic (potential bias toward higher shortage estimates)
Credibility rationaleData confirmed by Socialstyrelsen parallel report showing same regional trend
Wrong grade[B2] — ❌ Vårdförbundet has a sector interest that creates potential bias; reliability tier C is appropriate
Disambiguation noteIf Statskontoret or SCB independently publishes matching figure, upgrade to [B2]

Example 5 — Anonymous government source in tabloid

FieldValue
SourceAftonbladet citing "en källa nära statsministern" (unnamed, 2026-03-18)
Grade[D5]
Reliability rationaleTabloid reliability tier B for editorial standards, but unnamed source drops actual reliability to D (source cannot be evaluated)
Credibility rationaleClaim is that a specific ministerial resignation was imminent — contradicts public statements, inconsistent with pattern; improbable
Wrong grade[B4] — ❌ The reliability grade is for the source (unnamed tipster), not the publication; unnamed sources default to D
Disambiguation noteIf subsequent events confirm the tip, upgrade credibility to 1 but note the original reliability grade was D

Example 6 — Lagrådet official yttrande

FieldValue
SourceLagrådet yttrande Dnr 2026-031 (official PDF from lagradet.se)
Grade[A1]
Reliability rationaleOfficial primary-source document from an independent constitutional body; zero history of falsification
Credibility rationaleDocument text is the authoritative record; inherently confirmed
Disambiguation noteAnalyst's interpretation of whether the yttrande is "critical" (tillstyrker vs. kritiska anmärkningar) is a separate analytical judgment at [A3]

Example 7 — Social media post from party leader

FieldValue
SourceJimmie Åkesson (SD party leader) post on X/Twitter, 2026-04-15
Grade[B3]
Reliability rationaleVerified account of named political actor; no history of fake accounts; but posts are not fact-checked and may be political messaging
Credibility rationaleStatement of political position is plausible and consistent with party line; not independently corroborated
Wrong grade[A2] — ❌ Social media posts do not carry official publication standards; reliability tier B is ceiling for verified social media
Disambiguation noteIf the post is later reported verbatim by TT, the TT article carries [B2] and the original post remains [B3]

Example 8 — SCB housing price index (BO0501)

FieldValue
SourceSCB BO0501 — monthly house-price index, March 2026
Grade[A1]
Reliability rationaleOfficial national statistics; no history of revision errors; mandated public-interest publication
Credibility rationaleReleased simultaneously by SCB and confirmed by Riksbank housing-price tracker
Disambiguation noteFor model projections derived from SCB data (e.g. forecast for Q3 2026), grade the model separately at [B2]

Example 9 — Academic article on coalition durability

FieldValue
SourceBergman, T. et al. (2025) "Nordic coalition survival determinants", Scandinavian Political Studies
Grade[B2]
Reliability rationalePeer-reviewed journal; named researchers; no identified competing interest
Credibility rationaleFindings on Nordic coalition survival rates are consistent with independent Comparative Constitutions Project data
Wrong grade[A2] — ❌ Academic articles are not official government records; reliability tier B is appropriate for peer-reviewed academic sources

Example 10 — NATO official communiqué

FieldValue
SourceNATO Secretary-General press statement, 2026-03-22, on Swedish defence commitment
Grade[A2]
Reliability rationaleOfficial international organisation; primary source; no history of fabrication
Credibility rationaleStatement is the official NATO position — cannot be "confirmed" by another source independently
Disambiguation noteAnalyst interpretation of whether the statement implies a specific obligation for Sweden is a separate claim at [B3]

Example 11 — Party congress resolution (publicly available)

FieldValue
SourceSD party congress motion (publicly available PDF, 2025 congress)
Grade[B2]
Reliability rationaleOfficial party document; published by the party; verified by cross-check with SVT congress coverage
Credibility rationaleContent confirmed by SVT reporting and TT summary
Wrong grade[C3] — ❌ An official published party congress document is a B-grade source; C is for less institutional sources

Example 12 — EU Commission press release

FieldValue
SourceEuropean Commission DG Competition press release on state-aid investigation, 2026-05
Grade[A1]
Reliability rationaleOfficial EU institution publication; primary source; legally binding
Credibility rationaleCommission press release is authoritative confirmation of its own action

Example 13 — Riksbank minutes

FieldValue
SourceRiksbank monetary policy meeting minutes, April 2026
Grade[A2]
Reliability rationaleOfficial central bank publication; highest reliability tier
Credibility rationaleMinutes represent internal deliberation — content is Riksbank's own record, not externally corroborated

Example 14 — Swedish think-tank with transparent methodology

FieldValue
SourceSNS (Studieförbundet Näringsliv och Samhälle) housing report (2025)
Grade[C2]
Reliability rationaleNamed think-tank; published methodology; slight pro-market orientation creates potential bias → tier C
Credibility rationaleKey finding corroborated by Boverket official housing supply data
Wrong grade[B2] — ❌ SNS's known ideological orientation places it in reliability tier C despite transparent methodology

Example 15 — Unverified leaked document

FieldValue
SourcePurported Finansdepartementet internal memo (leaked to journalist, not confirmed by ministry)
Grade[F4]
Reliability rationaleSource (person who leaked) is unknown; document authenticity unverified → reliability F
Credibility rationaleContent is internally consistent with known fiscal trajectory but contradicts one official statement → doubtful
Wrong grade[C4] — ❌ An unverified document from an unknown leaker is an F-grade source regardless of content plausibility

Example 16 — Interpellation transcript (verbatim)

FieldValue
Sourceriksdagen.se interpellation transcript — verbatim ministerial answer 2026-05-03
Grade[A1]
Reliability rationaleOfficial Riksdag verbatim chamber record
Credibility rationaleDirect transcription; primary source; confirmed
Disambiguation noteAnalyst's quality-tier rating of the answer (tier 1–5 rubric) is a secondary analytical judgment at [A3] minimum

Example 17 — MSB threat assessment

FieldValue
SourceMSB (Myndigheten för samhällsskydd och beredskap) annual threat assessment (2025)
Grade[A2]
Reliability rationaleOfficial government agency; mandated threat assessment function; established track record
Credibility rationaleFindings consistent with SÄPO annual report trend but not independently confirmed by non-government source

Example 18 — Opposition motion (verbatim)

FieldValue
SourceRiksdag motion 2025/26:S482 (V party motion on housing) — verbatim text
Grade[A1]
Reliability rationaleOfficial registered Riksdag document; digital ledger record
Credibility rationaleVerbatim motion text; confirmed by Riksdag publication
Disambiguation noteAnalysis of the motion's likely outcome is a secondary claim at [B3] until vote records confirm

Example 19 — SVT opinion poll aggregation

FieldValue
SourceSVT opinionsmätning (Demoskop, March 2026) — published party support estimates
Grade[B2]
Reliability rationaleSVT public broadcaster; Demoskop established polling firm; transparent methodology; no history of systematic fabrication
Credibility rationaleConsistent with Sifo March 2026 (within margin of error)
Wrong grade[C3] — ❌ Demoskop has a long track record and SVT provides independent editorial oversight; tier B is appropriate
Disambiguation noteSingle poll without cross-poll corroboration → [B3]; confirmed by second major poll → [B2]

Example 20 — Parliamentary committee report

FieldValue
SourceFiU betänkande 2025/26:FiU28 (official Riksdag publication)
Grade[A1]
Reliability rationaleOfficial Riksdag committee report; legislative authority record
Credibility rationaleReport is the committee's authoritative output; confirmed by Riksdag publication
Disambiguation noteAnalyst predictions about how the report will vote in kammaren are secondary claims at [B3]

Example 21 — Statskontoret evaluation

FieldValue
SourceStatskontoret evaluation report on polisreformen (2025) — statskontoret.se
Grade[A2]
Reliability rationaleOfficial government expert body; independent mandate; primary data collection
Credibility rationaleFindings consistent with Riksrevisionen audit, but not all metrics independently confirmed
Disambiguation noteWhen both Statskontoret and Riksrevisionen independently reach the same finding, credibility upgrades to 1

Example 22 — ECtHR judgment

FieldValue
SourceEuropean Court of Human Rights judgment (Case Sweden, 2025) — hudoc.echr.coe.int
Grade[A1]
Reliability rationaleOfficial international court judgment; unimpeachable primary source
Credibility rationaleCourt judgment is confirmed primary record

⚠️ Common grading errors — do not repeat

Error patternCorrect approach
Applying the publication's grade to an unnamed source within itGrade the actual source (named person, named document); unnamed sources → D
Grading a secondary interpretation at the same level as the primary recordPrimary record (e.g. vote) = [A1]; analyst's interpretation of that record = [A3] at best
Upgrading a C source to B because its conclusions seem reasonableReasonableness is a credibility factor (1–6), not a reliability factor (A–F); keep reliability at C
Treating "two articles from the same newsroom" as two independent sourcesSame editorial chain = single source; credibility 1 requires sources from different editorial/institutional chains
Grading an AI model outputAI models are not intelligence sources; do not apply Admiralty grades to AI-generated content; only apply to primary sources cited by the AI

🧭 Inter-analyst calibration protocol

To maintain ≥80 % agreement between two analysts grading the same source:

  1. Grade independently before discussing.
  2. Check reliability first (letter grade): does the source type match the decision trees?
  3. Check credibility second (number grade): is there independent corroboration, and how consistent is the content?
  4. Disagreement resolution: if grades differ by ≥1 step (e.g. B vs. D, or 2 vs. 4), discuss using the worked-example sheet. The lower (more cautious) grade is adopted until resolution.
  5. Record agreement: log the final grade and any disagreement resolution note in methodology-reflection.md §Source grading audit.

Target: ≥80 % exact-grade agreement; ≤5 % two-step disagreements.



Document Control

  • Path: analysis/methodologies/admiralty-rubric.md
  • Classification: Public
  • Next Review: 2026-08-14