The Lightkeepers™ — Judges the Institutions Never Scored
A verified gallery of the humans at the end of warning chains: Petrov's kin. Dharmasaroja branded a scaremonger before the 2004 tsunami; the Challenger…
The Lightkeepers™ — judges the institutions never scored, honored AND graded
Every warning system ends in a human deciding believe-or-dismiss, and no institution keeps score on that human. This gallery holds the verified stories of Petrov's kin — each shipped with its documented core AND its owned correction, because the legends have accumulated flourishes the archives don't support, and a record whose brand is claims-that-survive-checking is the place where even the myths get checked. A Lightkeeper™ (the designation from the operator's book; trademark filing underway) is NOT a superforecaster, and the difference is the definition: superforecasting measures accuracy across many askable questions — a real and honorable standard this record submits to on every page — but no tournament could have scheduled Petrov's question (are these five missiles real, you have minutes); the moments that matter most cannot be pre-registered by any platform. The Lightkeeper standard weighs the other axis: the VALUE created at the moment of consequence — what the miss would have cost, what the judgment saved. Accuracy is the entry fee; consequence is the title. THE SCREEN TEST: apply the superforecaster screen (many pre-registered questions, tournament format, volume calibration) to every judge in this hall — all twenty-one FAIL, and the one who looks like a superforecaster (Finley, volume + 96.6%) is here precisely because Gilbert showed the number was base-rate-inflated. No tournament could have surfaced a single judge in the hall; their questions were unschedulable. Not a flaw in superforecasting — the boundary of it; the Lightkeeper™ standard begins where the schedule ends. The designation is earned by discipline under an unforgiving standard, never self-declared.
The verified eight: Smith Dharmasaroja (branded a scaremonger for tsunami warnings in 1993/1998, handed Thailand's National Disaster Warning Center after Boxing Day 2004 — punished on base rates, paid off eleven years later). Boisjoly & McDonald (Challenger eve: no-launch below 53°F reversed in a managers-only caucus — 'engineering hat' line attributed as sworn testimony; Rogers found flawed process, not malice). Sgt. John Finley & the tornado ban (1884: 96.6% claimed accuracy demolished by Gilbert's observation that 'no tornado' forever scores 98.2% — the exchange that FOUNDED forecast verification, direct ancestor of this record's Brier score; the word 'tornado' banned from forecasts 1887 until the modern era, 1948-1950). Frances Kelsey — THE COUNTER-PETROV, the judge who WAS scored: refused thalidomide approval 1960-62, presidential honor, Kefauver-Harris law (correction owned: ~17 US cases via investigational samples — the US was not fully spared). Hurricane Pam (the FEMA game that flooded a simulated New Orleans 13 months before Katrina; follow-through died 'due to funding' per sworn testimony — graded honestly: right on flood and displacement, off by an order of magnitude on deaths, missed the breach mode). Harry Markopolos (SEC OIG-509: repeated Madoff submissions 2000-2008, 'never properly examined' — the IG found process failure, explicitly NOT corruption). Joseph Rochefort (HYPO's Midway call backed by Nimitz over Washington; medal denied in life, DSM posthumous 1986 — and the widely repeated Presidential Medal of Freedom claim is FALSE, checked against the Reagan Library). John Snow (the pump handle came off AFTER the outbreak was declining — Snow said so himself; the establishment rejected him in 1855; the discipline, not the drama, is why he stands here).
The corrected tellings (tier two): Rescorla (with Dan Hill — 'warned of' aircraft attack per posthumous accounts, defied the stay-put order, 13 did not get out), the Yom Kippur konseptsia (institutional cognition, not malice; Marwan: fall, open verdict, contested loyalties — exactly that and no more), Semmelweis (died of sepsis; his own conduct contributed), Marshall & Warren (a decade of normal gatekeeping, then the Nobel), Rodney Rocha/Columbia (CAIB: organizational causes — no villains), the Phoenix memo & Rowley (blocked at FBI HQ, never reached the FISA court; the memo did NOT predict 9/11), Van Riper/MC02 (the command's own report: free play 'constrained to the point where the end state was scripted' — and the sinking was partly a sim artifact, owned), Billy Mitchell (court-martialed for 1925 insubordination, NOT for the 1924 Ford-Island-at-7:30am prediction — which is genuine; the real attack came at 7:55), Wegener (right pattern, wrong mechanism, wrong pedigree), Woodford/Olympus (suspended sentences — 'executives went to prison' is largely false).
Attested, not corroborated (tier three — the flourishes, graded): Arkhipov/B-59 (the celebrated intervention rests on one participant's 2002 oral account; Arkhipov's own 1997 talk omits it — attested by testimony, not corroborated by documents), Able Archer 83 (the PFIAB review said Soviet alarm was real but its depth 'difficult to gauge'; the near-launch telling is not established), Sorge (the 'named June 22' telegram is a later Soviet embellishment; closest verified: 'latter part of June'). These stay in the gallery because the documented layers matter; they stay in this section because this site does not promote testimony to fact.
What the gallery is for: twenty-one stories, one pattern — the judgment layer runs unscored and the people who were right paid for it. The fix costs a hash and a timestamp. This record is the working demonstration: every call sealed before its event, graded in public, misses kept at full weight (/dashboard, /grading, /verify). The standing exhibit — Stanislav Petrov, the king of the class — presides over the gallery itself (full signature panel on the page) and opens /warfighter. Richard Feynman presides as PROFESSOR EMERITUS — the judge of judges: on the Rogers Commission he refused testimony as a substitute for demonstration (the televised O-ring ice-water exhibit) and wrote the sentence this record is built under: 'For a successful technology, reality must take precedence over public relations, for nature cannot be fooled' (Appendix F). Related: /wargaming (Hurricane Pam seed), /space-force (Challenger seed).