{"version":"0.1.0","as_of":"2026-05-19","rationale":"Per-category trust priors used by the signal-analyzer to weight source claims when corroboration is absent. These are PRIORS, not final weights — the empirical /source-quality ablation overrides them once enough data accumulates. The prior is the probability we'd assign to an isolated claim from this category being correct, in the absence of other signals. Categories with prior < 0.5 are treated as 'adversarial': their loud claims about themselves are discounted, but their concessions (admitting own setbacks) are upweighted because the natural incentive is to suppress them.","categories":{"WIRE":{"prior":0.95,"examples":["Reuters","Associated Press","AFP","BBC"],"stance":"neutral","notes":"Bottom-line factual, multi-source corroborated, low partisan distortion. Used as ground-truth baseline for cross-checking other categories."},"OFFICIAL_GOV_ALIGNED":{"prior":0.7,"examples":["US State Dept","White House","IDF spokesperson","UK FCDO","EU External Action"],"stance":"aligned_west","notes":"Factual on own actions/policy; framing is biased toward national interest. Statements ABOUT adversary actions need cross-checking. Statements OF policy intent are gold-standard signal — they ARE the policy."},"OFFICIAL_GOV_ADVERSARIAL":{"prior":0.55,"examples":["IRGC","Iranian MFA","Russian MFA","Hamas military wing"],"stance":"adversarial","notes":"Self-actions framed as deliberate policy carry HIGH signal even if factually distorted; the political choice to publish at all is informative. Confirmation of own setback (e.g. IRGC admitting a strike landed) is very strong because suppression would be the default. Inflated claims about adversary losses get inverse-updated."},"STATE_MEDIA_ALLIED":{"prior":0.65,"examples":["Times of Israel","Ynet","Kyiv Post","Al Arabiya English"],"stance":"aligned","notes":"Reports closer to events than wires (Hebrew-language outlets read IDF press briefings live). Editorial slant pro-establishment. Generally accurate on hard facts; framing tilted."},"STATE_MEDIA_ADVERSARIAL":{"prior":0.4,"examples":["TASS","RIA Novosti","IRNA","PressTV","Xinhua","Mehr"],"stance":"adversarial","notes":"Direct propaganda organs. Treat as adversarial: claims about own state's actions are framed; claims about adversary losses are routinely inflated. The inverse-update rule: a Russian state outlet admitting a Russian setback is high-confidence true (the incentive is to suppress)."},"OSINT_INDEPENDENT":{"prior":0.85,"examples":["Bellingcat","Aurora Intel","ME Spectator","Nuclear Iran","Flash Point ME"],"stance":"neutral","notes":"Independent analyst accounts with verifiable methodologies (sat imagery, geolocation, ADS-B). Highest trust for real-time tactical events. Some risk of repost-without-verification in fast-moving situations; cross-check time-sensitive claims."},"OSINT_AFFILIATED":{"prior":0.6,"examples":["Ukraine Weapons Tracker","InformNapalm"],"stance":"aligned","notes":"Affiliated OSINT with documented partisan lean. Strong on equipment-level forensics; partial on casualty/morale claims. Discount per-side claims accordingly."},"AGGREGATOR":{"prior":0.5,"examples":["Google News","Manifold consensus","Metaculus","Twitter trending"],"stance":"neutral","notes":"Derivative — surfaces crowd consensus, not ground truth. Useful for measuring market-implied probability, not for primary signal. Crowd is wisdom of, but also crowd of, etc."},"WHALE_FLOW":{"prior":0.75,"examples":["Polymarket smart-money trades","Kalshi taker volume","Whale Watcher tweets"],"stance":"skin_in_game","notes":"Real money, real cost-of-being-wrong. High informational value — these accounts have explicit P&L incentives to be right. But beware whale concentration: 1-2 actors can dominate flow without representing consensus."},"FINANCIAL_INDICATOR":{"prior":0.9,"examples":["Brent crude","Gold","VIX","ILS/USD","Defense ETF (ITA)"],"stance":"skin_in_game","notes":"Multi-trillion-dollar markets pricing risk in real time. Very hard to fake or manipulate at scale. Brent + ILS/USD especially load on Iran/Israel theater. Caveat: macro spillover (Fed, oil supply) can move these without geopolitical cause."},"CYBER_FEED":{"prior":0.85,"examples":["CISA Advisories","Microsoft Security","Mandiant","BleepingComputer","The Record (Recorded Future)"],"stance":"neutral","notes":"High trust for cyber-domain events (APT campaigns, CVE exploits). Lower trust for cross-domain (cyber events as indicators of geopolitical escalation are derivative; treat as weak corroboration not standalone signal)."},"PARTISAN_COMMENTARY":{"prior":0.4,"examples":["FDD","Quincy Institute","IRGC-aligned commentary accounts","advocacy think tanks"],"stance":"partisan","notes":"Strong priors color factual claims. Useful as a sentiment-of-the-camp signal (what does the hawk side think today?) not as ground truth. Cite for framing, never as primary fact."},"SEISMIC":{"prior":0.95,"examples":["USGS GeoJSON feed"],"stance":"neutral","notes":"Public scientific instrument with no political incentive. Filter is the trust question: is this Iran-region seismic event consistent with a nuclear test signature, or natural? Proximity-to-nuclear-site filter handles that."}},"update_policy":{"frequency":"These priors are seeded from domain knowledge and revised when the empirical /source-quality ablation produces a statistically meaningful disagreement (n>=30 paired trades per source-type, non-overlapping Wilson CIs).","fallback":"When a source's empirical track record is in the insufficient_data state, the prior is the operative weight. As n grows, the empirical signal replaces the prior smoothly via Bayesian update."}}