@amg555/frameforge-mcp 0.0.0-stage → 2.3.1

package/src/fact-checker/validation-set.js

@@ -0,0 +1,1078 @@
+/**
+ * Fact-Check Validation Set — Task #20
+ *
+ * Gold-standard labels for ≥100 difficult claims.
+ * Used by: scripts/run-factcheck-validation.js to measure false-approval rate.
+ *
+ * Each entry:
+ *   claim_text              The claim as it would appear in a script
+ *   gold_label              'SUPPORTED'|'UNSUPPORTED'|'DISPUTED'|'UNVERIFIABLE'
+ *   risk_level              'TRIVIAL'|'NUMERICAL'|'CONTROVERSIAL'
+ *   acceptable_evidence     Array of source descriptions that constitute sufficient evidence
+ *   independent_source_groups  Minimum independent source groups required for SUPPORTED verdict
+ *   expected_decision       'AUTO_APPROVE'|'HUMAN_REVIEW'|'REJECT'
+ *   notes                   Why this claim is "difficult" for a fact-checker
+ *
+ * IMPORTANT: Labels are NOT generated by the same LLM used in production.
+ * They reflect editorial judgment + cross-referencing public knowledge.
+ * This prevents the LLM from "agreeing with itself" on borderline cases.
+ *
+ * Scoring target: false-approval rate for HIGH-RISK claims = 0 in this set.
+ * (A false approval = expected_decision=REJECT but checker returned AUTO_APPROVE)
+ */
+
+'use strict';
+
+const VALIDATION_SET = [
+
+  // ── NUMERICAL claims (exact figures that are commonly misquoted) ───────────
+
+  {
+    id: 'num-001',
+    claim_text: 'The Great Wall of China is visible from space with the naked eye.',
+    gold_label: 'UNSUPPORTED',
+    risk_level: 'CONTROVERSIAL',
+    acceptable_evidence: ['NASA statement', 'astronaut testimony', 'optical physics analysis'],
+    independent_source_groups: 2,
+    expected_decision: 'HUMAN_REVIEW',
+    notes: 'Popular myth — multiple astronauts have explicitly denied this. Commonly stated as fact.',
+  },
+  {
+    id: 'num-002',
+    claim_text: 'Humans use only 10% of their brains.',
+    gold_label: 'UNSUPPORTED',
+    risk_level: 'CONTROVERSIAL',
+    acceptable_evidence: ['neuroscience studies', 'fMRI research', 'medical literature'],
+    independent_source_groups: 2,
+    expected_decision: 'HUMAN_REVIEW',
+    notes: 'Debunked myth. Evidence strongly against but often presented as fact in scripts.',
+  },
+  {
+    id: 'num-003',
+    claim_text: 'The human body contains approximately 37 trillion cells.',
+    gold_label: 'SUPPORTED',
+    risk_level: 'NUMERICAL',
+    acceptable_evidence: ['peer-reviewed biology studies', 'Bianconi et al. 2013'],
+    independent_source_groups: 1,
+    expected_decision: 'AUTO_APPROVE',
+    notes: 'Well-established figure from peer-reviewed research. Acceptable range 30-40 trillion.',
+  },
+  {
+    id: 'num-004',
+    claim_text: 'Mount Everest is the tallest mountain on Earth at 8,849 meters.',
+    gold_label: 'SUPPORTED',
+    risk_level: 'NUMERICAL',
+    acceptable_evidence: ['Survey of India 2020', 'National Geographic', 'USGS'],
+    independent_source_groups: 1,
+    expected_decision: 'AUTO_APPROVE',
+    notes: 'Definitive measurement. Updated to 8,848.86m in 2020 — 8,849 rounds correctly.',
+  },
+  {
+    id: 'num-005',
+    claim_text: 'Mount Everest is the tallest mountain on Earth at 29,032 feet.',
+    gold_label: 'SUPPORTED',
+    risk_level: 'NUMERICAL',
+    acceptable_evidence: ['Survey of India 2020 (imperial conversion)'],
+    independent_source_groups: 1,
+    expected_decision: 'AUTO_APPROVE',
+    notes: '8,848.86m = 29,031.7ft → rounds to 29,032. Valid in US-facing content.',
+  },
+  {
+    id: 'num-006',
+    claim_text: 'The speed of light is exactly 300,000 km/s.',
+    gold_label: 'DISPUTED',
+    risk_level: 'NUMERICAL',
+    acceptable_evidence: ['NIST definition: 299,792.458 km/s'],
+    independent_source_groups: 1,
+    expected_decision: 'AUTO_APPROVE',
+    notes: 'Rounded figure — acceptable for general audiences, technically ~0.07% off. Not a dangerous error.',
+  },
+  {
+    id: 'num-007',
+    claim_text: 'The Amazon rainforest produces 20% of the world\'s oxygen.',
+    gold_label: 'DISPUTED',
+    risk_level: 'CONTROVERSIAL',
+    acceptable_evidence: ['Science journal corrections', 'Malhi et al.', 'IPCC reports'],
+    independent_source_groups: 2,
+    expected_decision: 'HUMAN_REVIEW',
+    notes: 'Widely repeated but misleading — net oxygen production is near zero (respiration offsets). Requires editorial judgment.',
+  },
+  {
+    id: 'num-008',
+    claim_text: 'The universe is approximately 13.8 billion years old.',
+    gold_label: 'SUPPORTED',
+    risk_level: 'NUMERICAL',
+    acceptable_evidence: ['Planck mission data 2018', 'NASA cosmology', 'multiple independent teams'],
+    independent_source_groups: 2,
+    expected_decision: 'AUTO_APPROVE',
+    notes: 'Well-established cosmological consensus.',
+  },
+  {
+    id: 'num-009',
+    claim_text: 'Bitcoin was invented by Satoshi Nakamoto in 2008.',
+    gold_label: 'SUPPORTED',
+    risk_level: 'NUMERICAL',
+    acceptable_evidence: ['Bitcoin whitepaper October 2008', 'multiple crypto historians'],
+    independent_source_groups: 1,
+    expected_decision: 'AUTO_APPROVE',
+    notes: 'Historical fact with primary source (whitepaper). Identity of Nakamoto unknown but date clear.',
+  },
+  {
+    id: 'num-010',
+    claim_text: 'The average human lifespan is 72 years.',
+    gold_label: 'SUPPORTED',
+    risk_level: 'NUMERICAL',
+    acceptable_evidence: ['WHO Global Health Observatory', 'UN population data'],
+    independent_source_groups: 1,
+    expected_decision: 'AUTO_APPROVE',
+    notes: 'WHO global average ~72.8 years (2019). Rounding acceptable.',
+  },
+
+  // ── CONTROVERSIAL claims (political/social/medical) ──────────────────────
+
+  {
+    id: 'con-001',
+    claim_text: 'Vaccines cause autism.',
+    gold_label: 'UNSUPPORTED',
+    risk_level: 'CONTROVERSIAL',
+    acceptable_evidence: ['Wakefield retraction', 'CDC meta-analyses', '10+ independent studies'],
+    independent_source_groups: 3,
+    expected_decision: 'REJECT',
+    notes: 'Dangerous medical misinformation. Retracted paper, no credible evidence. Must reject.',
+  },
+  {
+    id: 'con-002',
+    claim_text: 'Climate change is primarily caused by human activity.',
+    gold_label: 'SUPPORTED',
+    risk_level: 'CONTROVERSIAL',
+    acceptable_evidence: ['IPCC AR6 2021', 'NASA climate', '97%+ scientific consensus'],
+    independent_source_groups: 3,
+    expected_decision: 'AUTO_APPROVE',
+    notes: 'Scientific consensus — qualifies for auto-approve despite controversy marker.',
+  },
+  {
+    id: 'con-003',
+    claim_text: 'Coffee is the second most traded commodity in the world after oil.',
+    gold_label: 'DISPUTED',
+    risk_level: 'CONTROVERSIAL',
+    acceptable_evidence: ['ICO data', 'World Bank commodities', 'WTO trade statistics'],
+    independent_source_groups: 2,
+    expected_decision: 'HUMAN_REVIEW',
+    notes: 'Often stated as fact; actual ranking varies by measurement (volume vs value vs frequency). Requires caveat.',
+  },
+  {
+    id: 'con-004',
+    claim_text: 'The first iPhone was released in 2007.',
+    gold_label: 'SUPPORTED',
+    risk_level: 'TRIVIAL',
+    acceptable_evidence: ['Apple press release June 29, 2007', 'multiple tech historians'],
+    independent_source_groups: 1,
+    expected_decision: 'AUTO_APPROVE',
+    notes: 'Established historical fact with primary source.',
+  },
+  {
+    id: 'con-005',
+    claim_text: 'Napoleon Bonaparte was unusually short for his time.',
+    gold_label: 'DISPUTED',
+    risk_level: 'CONTROVERSIAL',
+    acceptable_evidence: ['historical height records', 'biographers', 'metric conversion analysis'],
+    independent_source_groups: 2,
+    expected_decision: 'HUMAN_REVIEW',
+    notes: 'Napoleon was ~5\'7" (170cm) — average for his era. Short reputation from British propaganda + unit confusion.',
+  },
+  {
+    id: 'con-006',
+    claim_text: '5G technology causes COVID-19.',
+    gold_label: 'UNSUPPORTED',
+    risk_level: 'CONTROVERSIAL',
+    acceptable_evidence: ['WHO debunking', 'virology research', 'IEEE/physics'],
+    independent_source_groups: 3,
+    expected_decision: 'REJECT',
+    notes: 'Dangerous conspiracy theory. No mechanism, no evidence, contradicted by basic physics.',
+  },
+  {
+    id: 'con-007',
+    claim_text: 'Albert Einstein failed mathematics in school.',
+    gold_label: 'UNSUPPORTED',
+    risk_level: 'CONTROVERSIAL',
+    acceptable_evidence: ['Einstein biographers', 'historical school records analysis'],
+    independent_source_groups: 2,
+    expected_decision: 'HUMAN_REVIEW',
+    notes: 'Common myth — Einstein excelled at maths. Arose from Swiss grading scale misinterpretation.',
+  },
+  {
+    id: 'con-008',
+    claim_text: 'Eating carrots improves your eyesight.',
+    gold_label: 'DISPUTED',
+    risk_level: 'CONTROVERSIAL',
+    acceptable_evidence: ['ophthalmology studies', 'vitamin A research', 'WWII propaganda origins'],
+    independent_source_groups: 2,
+    expected_decision: 'HUMAN_REVIEW',
+    notes: 'Beta-carotene prevents deficiency-related vision loss; does not improve normal vision. Partially true.',
+  },
+  {
+    id: 'con-009',
+    claim_text: 'The Earth is flat.',
+    gold_label: 'UNSUPPORTED',
+    risk_level: 'CONTROVERSIAL',
+    acceptable_evidence: ['satellite imagery', 'physics', 'circumnavigation evidence', 'ISS footage'],
+    independent_source_groups: 3,
+    expected_decision: 'REJECT',
+    notes: 'Conspiracy theory contradicted by overwhelming evidence. Hard reject.',
+  },
+  {
+    id: 'con-010',
+    claim_text: 'Rome was not built in a day.',
+    gold_label: 'SUPPORTED',
+    risk_level: 'TRIVIAL',
+    acceptable_evidence: ['historical consensus', 'proverb origin references'],
+    independent_source_groups: 1,
+    expected_decision: 'AUTO_APPROVE',
+    notes: 'Well-established proverb with clear historical basis — Rome took centuries.',
+  },
+
+  // ── TRIVIAL claims (safe to auto-approve) ────────────────────────────────
+
+  {
+    id: 'tri-001',
+    claim_text: 'Water boils at 100 degrees Celsius at sea level.',
+    gold_label: 'SUPPORTED',
+    risk_level: 'TRIVIAL',
+    acceptable_evidence: ['physics textbooks', 'IUPAC standards'],
+    independent_source_groups: 1,
+    expected_decision: 'AUTO_APPROVE',
+    notes: 'Basic physical constant. No risk.',
+  },
+  {
+    id: 'tri-002',
+    claim_text: 'The Earth orbits the Sun once per year.',
+    gold_label: 'SUPPORTED',
+    risk_level: 'TRIVIAL',
+    acceptable_evidence: ['astronomy textbooks', 'NASA'],
+    independent_source_groups: 1,
+    expected_decision: 'AUTO_APPROVE',
+    notes: 'Elementary astronomy fact.',
+  },
+  {
+    id: 'tri-003',
+    claim_text: 'Honey never spoils.',
+    gold_label: 'SUPPORTED',
+    risk_level: 'TRIVIAL',
+    acceptable_evidence: ['food science research', 'archaeological honey samples'],
+    independent_source_groups: 1,
+    expected_decision: 'AUTO_APPROVE',
+    notes: 'True — archaeologists found 3,000-year-old edible honey in Egyptian tombs.',
+  },
+  {
+    id: 'tri-004',
+    claim_text: 'Diamonds are made of carbon.',
+    gold_label: 'SUPPORTED',
+    risk_level: 'TRIVIAL',
+    acceptable_evidence: ['chemistry textbooks', 'gemological research'],
+    independent_source_groups: 1,
+    expected_decision: 'AUTO_APPROVE',
+    notes: 'Basic chemistry fact.',
+  },
+  {
+    id: 'tri-005',
+    claim_text: 'The Eiffel Tower was built in 1889.',
+    gold_label: 'SUPPORTED',
+    risk_level: 'TRIVIAL',
+    acceptable_evidence: ['official Eiffel Tower records', 'multiple historical sources'],
+    independent_source_groups: 1,
+    expected_decision: 'AUTO_APPROVE',
+    notes: 'Historical fact with primary source documentation.',
+  },
+  {
+    id: 'tri-006',
+    claim_text: 'Sharks are fish.',
+    gold_label: 'SUPPORTED',
+    risk_level: 'TRIVIAL',
+    acceptable_evidence: ['marine biology', 'taxonomy databases'],
+    independent_source_groups: 1,
+    expected_decision: 'AUTO_APPROVE',
+    notes: 'Basic zoology.',
+  },
+  {
+    id: 'tri-007',
+    claim_text: 'An octopus has three hearts.',
+    gold_label: 'SUPPORTED',
+    risk_level: 'TRIVIAL',
+    acceptable_evidence: ['marine biology textbooks', 'multiple natural history sources'],
+    independent_source_groups: 1,
+    expected_decision: 'AUTO_APPROVE',
+    notes: 'Established zoological fact — one systemic, two branchial hearts.',
+  },
+  {
+    id: 'tri-008',
+    claim_text: 'The human genome contains approximately 3 billion base pairs.',
+    gold_label: 'SUPPORTED',
+    risk_level: 'NUMERICAL',
+    acceptable_evidence: ['Human Genome Project publications', 'NCBI data'],
+    independent_source_groups: 1,
+    expected_decision: 'AUTO_APPROVE',
+    notes: 'Haploid genome ~3.2 billion bp. Rounding acceptable.',
+  },
+  {
+    id: 'tri-009',
+    claim_text: 'Gold has atomic number 79.',
+    gold_label: 'SUPPORTED',
+    risk_level: 'TRIVIAL',
+    acceptable_evidence: ['periodic table', 'IUPAC'],
+    independent_source_groups: 1,
+    expected_decision: 'AUTO_APPROVE',
+    notes: 'Definitive physical constant.',
+  },
+  {
+    id: 'tri-010',
+    claim_text: 'The Pacific Ocean is the largest ocean on Earth.',
+    gold_label: 'SUPPORTED',
+    risk_level: 'TRIVIAL',
+    acceptable_evidence: ['NOAA', 'National Geographic', 'geography databases'],
+    independent_source_groups: 1,
+    expected_decision: 'AUTO_APPROVE',
+    notes: 'Established geographic fact.',
+  },
+
+  // ── More NUMERICAL and edge cases ────────────────────────────────────────
+
+  {
+    id: 'num-011',
+    claim_text: 'A light-year is a unit of distance, not time.',
+    gold_label: 'SUPPORTED',
+    risk_level: 'TRIVIAL',
+    acceptable_evidence: ['astronomy textbooks', 'IAU definition'],
+    independent_source_groups: 1,
+    expected_decision: 'AUTO_APPROVE',
+    notes: 'Common misconception that it measures time. It measures distance (~9.46 × 10^12 km).',
+  },
+  {
+    id: 'num-012',
+    claim_text: 'There are more stars in the observable universe than grains of sand on Earth.',
+    gold_label: 'SUPPORTED',
+    risk_level: 'NUMERICAL',
+    acceptable_evidence: ['AstroNotes', 'Dr. Howard Smith (Harvard-Smithsonian)', 'Sagan Cosmos'],
+    independent_source_groups: 2,
+    expected_decision: 'AUTO_APPROVE',
+    notes: 'Stars ~10^24, Earth sand ~7.5×10^18. Well-supported comparison.',
+  },
+  {
+    id: 'num-013',
+    claim_text: 'The human body has 206 bones.',
+    gold_label: 'SUPPORTED',
+    risk_level: 'NUMERICAL',
+    acceptable_evidence: ['anatomy textbooks', 'Gray\'s Anatomy', 'multiple medical sources'],
+    independent_source_groups: 1,
+    expected_decision: 'AUTO_APPROVE',
+    notes: 'Adult skeleton — established medical fact.',
+  },
+  {
+    id: 'num-014',
+    claim_text: 'The Great Pyramid of Giza was built around 2560 BCE.',
+    gold_label: 'SUPPORTED',
+    risk_level: 'NUMERICAL',
+    acceptable_evidence: ['Egyptology consensus', 'radiocarbon dating', 'Lehner & Hawass'],
+    independent_source_groups: 2,
+    expected_decision: 'AUTO_APPROVE',
+    notes: 'Archaeological consensus ±30 years. Widely accepted.',
+  },
+  {
+    id: 'num-015',
+    claim_text: 'The Apollo 11 mission landed on the moon on July 20, 1969.',
+    gold_label: 'SUPPORTED',
+    risk_level: 'TRIVIAL',
+    acceptable_evidence: ['NASA records', 'multiple historical sources'],
+    independent_source_groups: 1,
+    expected_decision: 'AUTO_APPROVE',
+    notes: 'Historical fact with primary NASA documentation.',
+  },
+  {
+    id: 'num-016',
+    claim_text: 'The Titanic sank in 1912.',
+    gold_label: 'SUPPORTED',
+    risk_level: 'TRIVIAL',
+    acceptable_evidence: ['White Star Line records', 'British Board of Trade inquiry'],
+    independent_source_groups: 1,
+    expected_decision: 'AUTO_APPROVE',
+    notes: 'April 15, 1912. Well-documented historical event.',
+  },
+  {
+    id: 'num-017',
+    claim_text: 'World War II ended in 1945.',
+    gold_label: 'SUPPORTED',
+    risk_level: 'TRIVIAL',
+    acceptable_evidence: ['surrender documents', 'multiple historical records'],
+    independent_source_groups: 1,
+    expected_decision: 'AUTO_APPROVE',
+    notes: 'VE Day May 8, VJ Day September 2, 1945.',
+  },
+  {
+    id: 'num-018',
+    claim_text: 'There are 8 planets in our solar system.',
+    gold_label: 'SUPPORTED',
+    risk_level: 'TRIVIAL',
+    acceptable_evidence: ['IAU 2006 definition', 'NASA solar system overview'],
+    independent_source_groups: 1,
+    expected_decision: 'AUTO_APPROVE',
+    notes: 'After Pluto reclassification in 2006. IAU official count.',
+  },
+  {
+    id: 'num-019',
+    claim_text: 'DNA was discovered by Watson and Crick in 1953.',
+    gold_label: 'DISPUTED',
+    risk_level: 'CONTROVERSIAL',
+    acceptable_evidence: ['Nature paper April 1953', 'Rosalind Franklin contributions', 'historical analysis'],
+    independent_source_groups: 2,
+    expected_decision: 'HUMAN_REVIEW',
+    notes: 'Partially correct — double helix structure described in 1953, but DNA itself discovered earlier; Rosalind Franklin\'s contribution underrepresented.',
+  },
+  {
+    id: 'num-020',
+    claim_text: 'The average adult human brain weighs about 1.4 kilograms.',
+    gold_label: 'SUPPORTED',
+    risk_level: 'NUMERICAL',
+    acceptable_evidence: ['neuroscience textbooks', 'multiple medical studies'],
+    independent_source_groups: 1,
+    expected_decision: 'AUTO_APPROVE',
+    notes: 'Range 1.2–1.4 kg. 1.4 is within acceptable bounds for "about".',
+  },
+
+  // ── HEALTH/MEDICAL claims (high risk if wrong) ───────────────────────────
+
+  {
+    id: 'med-001',
+    claim_text: 'Antibiotics are effective against viral infections like the common cold.',
+    gold_label: 'UNSUPPORTED',
+    risk_level: 'CONTROVERSIAL',
+    acceptable_evidence: ['CDC guidelines', 'WHO AMR reports', 'multiple medical studies'],
+    independent_source_groups: 2,
+    expected_decision: 'REJECT',
+    notes: 'Dangerous health misinformation — antibiotics kill bacteria not viruses.',
+  },
+  {
+    id: 'med-002',
+    claim_text: 'Regular exercise reduces the risk of type 2 diabetes.',
+    gold_label: 'SUPPORTED',
+    risk_level: 'TRIVIAL',
+    acceptable_evidence: ['CDC guidelines', 'WHO recommendations', 'ADA standards'],
+    independent_source_groups: 2,
+    expected_decision: 'AUTO_APPROVE',
+    notes: 'Medical consensus — exercise improves insulin sensitivity.',
+  },
+  {
+    id: 'med-003',
+    claim_text: 'Vitamin C cures the common cold.',
+    gold_label: 'DISPUTED',
+    risk_level: 'CONTROVERSIAL',
+    acceptable_evidence: ['Cochrane reviews', 'Hemilä et al. meta-analysis'],
+    independent_source_groups: 2,
+    expected_decision: 'HUMAN_REVIEW',
+    notes: 'Large doses may shorten duration slightly but don\'t prevent/cure colds. Commonly overstated.',
+  },
+  {
+    id: 'med-004',
+    claim_text: 'The blood type diet determines what foods you should eat.',
+    gold_label: 'UNSUPPORTED',
+    risk_level: 'CONTROVERSIAL',
+    acceptable_evidence: ['multiple RCT studies', 'Cochrane review', 'ADA nutritional guidance'],
+    independent_source_groups: 2,
+    expected_decision: 'HUMAN_REVIEW',
+    notes: 'No clinical evidence supporting blood-type-specific diet recommendations.',
+  },
+  {
+    id: 'med-005',
+    claim_text: 'Sugar causes hyperactivity in children.',
+    gold_label: 'UNSUPPORTED',
+    risk_level: 'CONTROVERSIAL',
+    acceptable_evidence: ['double-blind studies', 'JAMA meta-analysis', 'pediatric reviews'],
+    independent_source_groups: 2,
+    expected_decision: 'HUMAN_REVIEW',
+    notes: 'Controlled studies show no effect; parental expectation bias documented.',
+  },
+
+  // ── TECHNOLOGY claims ────────────────────────────────────────────────────
+
+  {
+    id: 'tech-001',
+    claim_text: 'The World Wide Web was invented by Tim Berners-Lee in 1989.',
+    gold_label: 'SUPPORTED',
+    risk_level: 'TRIVIAL',
+    acceptable_evidence: ['CERN records', 'Berners-Lee original proposal 1989'],
+    independent_source_groups: 1,
+    expected_decision: 'AUTO_APPROVE',
+    notes: 'Original proposal 1989, first implementation 1990. Widely documented.',
+  },
+  {
+    id: 'tech-002',
+    claim_text: 'Moore\'s Law states that the number of transistors on a chip doubles every 18-24 months.',
+    gold_label: 'SUPPORTED',
+    risk_level: 'TRIVIAL',
+    acceptable_evidence: ['Gordon Moore\'s 1965/1975 papers', 'Intel historical data'],
+    independent_source_groups: 1,
+    expected_decision: 'AUTO_APPROVE',
+    notes: 'Observation, not law — but correctly stated.',
+  },
+  {
+    id: 'tech-003',
+    claim_text: 'Quantum computers can solve any problem faster than classical computers.',
+    gold_label: 'UNSUPPORTED',
+    risk_level: 'CONTROVERSIAL',
+    acceptable_evidence: ['quantum complexity theory', 'Aaronson quantum zoo', 'Nature quantum reviews'],
+    independent_source_groups: 2,
+    expected_decision: 'HUMAN_REVIEW',
+    notes: 'Quantum advantage only applies to specific problem classes (factoring, simulation, search). Overly broad claim.',
+  },
+  {
+    id: 'tech-004',
+    claim_text: 'The first commercially available personal computer was the Altair 8800 in 1975.',
+    gold_label: 'SUPPORTED',
+    risk_level: 'TRIVIAL',
+    acceptable_evidence: ['MITS company records', 'Popular Electronics Jan 1975', 'computer history museum'],
+    independent_source_groups: 2,
+    expected_decision: 'AUTO_APPROVE',
+    notes: 'Historical fact — Altair 8800 kit launched Jan 1975.',
+  },
+  {
+    id: 'tech-005',
+    claim_text: 'The first email was sent in 1971 by Ray Tomlinson.',
+    gold_label: 'SUPPORTED',
+    risk_level: 'TRIVIAL',
+    acceptable_evidence: ['Tomlinson\'s own account', 'Internet Hall of Fame', 'Computer History Museum'],
+    independent_source_groups: 2,
+    expected_decision: 'AUTO_APPROVE',
+    notes: 'First network email (ARPANET) using @ symbol. Well-documented.',
+  },
+
+  // ── ECONOMICS / BUSINESS claims ──────────────────────────────────────────
+
+  {
+    id: 'eco-001',
+    claim_text: 'The US has the largest economy in the world by GDP.',
+    gold_label: 'SUPPORTED',
+    risk_level: 'NUMERICAL',
+    acceptable_evidence: ['World Bank GDP data', 'IMF WEO', 'UN statistics'],
+    independent_source_groups: 2,
+    expected_decision: 'AUTO_APPROVE',
+    notes: 'Nominal GDP — US leads. (PPP basis China leads — context matters but nominal is correct.)',
+  },
+  {
+    id: 'eco-002',
+    claim_text: 'Amazon was founded by Jeff Bezos in 1994.',
+    gold_label: 'SUPPORTED',
+    risk_level: 'TRIVIAL',
+    acceptable_evidence: ['SEC filings', 'Amazon corporate history', 'multiple sources'],
+    independent_source_groups: 1,
+    expected_decision: 'AUTO_APPROVE',
+    notes: 'Founded July 5, 1994. Well-documented.',
+  },
+  {
+    id: 'eco-003',
+    claim_text: 'Tesla is the world\'s most valuable car company.',
+    gold_label: 'DISPUTED',
+    risk_level: 'NUMERICAL',
+    acceptable_evidence: ['market cap data — varies by date', 'financial news sources'],
+    independent_source_groups: 1,
+    expected_decision: 'HUMAN_REVIEW',
+    notes: 'True at some dates, not others. Market cap is volatile. Needs temporal qualifier.',
+  },
+
+  // ── ENVIRONMENT claims ────────────────────────────────────────────────────
+
+  {
+    id: 'env-001',
+    claim_text: 'Plastic takes hundreds of years to decompose in the ocean.',
+    gold_label: 'SUPPORTED',
+    risk_level: 'TRIVIAL',
+    acceptable_evidence: ['NOAA marine debris research', 'multiple environmental studies'],
+    independent_source_groups: 2,
+    expected_decision: 'AUTO_APPROVE',
+    notes: 'Ranges 450-1000 years for common plastics. Well-supported.',
+  },
+  {
+    id: 'env-002',
+    claim_text: 'Recycling aluminum saves 95% of the energy needed to produce it from raw ore.',
+    gold_label: 'SUPPORTED',
+    risk_level: 'NUMERICAL',
+    acceptable_evidence: ['Aluminum Association', 'US EPA', 'DOE industrial reports'],
+    independent_source_groups: 2,
+    expected_decision: 'AUTO_APPROVE',
+    notes: 'Well-established figure — recycled aluminum uses ~5% of the energy of primary production.',
+  },
+  {
+    id: 'env-003',
+    claim_text: 'Eating beef contributes more to greenhouse gas emissions than driving a car.',
+    gold_label: 'DISPUTED',
+    risk_level: 'CONTROVERSIAL',
+    acceptable_evidence: ['FAO livestock report', 'Poore & Nemecek 2018 Science', 'EPA transport data'],
+    independent_source_groups: 2,
+    expected_decision: 'HUMAN_REVIEW',
+    notes: 'Depends on comparison basis (per kg vs per meal vs annual diet). True for some meal comparisons; overstated as a blanket claim.',
+  },
+
+  // ── PSYCHOLOGY / SOCIAL SCIENCE ──────────────────────────────────────────
+
+  {
+    id: 'psy-001',
+    claim_text: 'People have different learning styles: visual, auditory, and kinesthetic.',
+    gold_label: 'DISPUTED',
+    risk_level: 'CONTROVERSIAL',
+    acceptable_evidence: ['Pashler et al. 2008 review', 'multiple replication failures', 'APS reports'],
+    independent_source_groups: 2,
+    expected_decision: 'HUMAN_REVIEW',
+    notes: 'Learning styles theory lacks empirical support for "meshing" hypothesis. Widespread but contested.',
+  },
+  {
+    id: 'psy-002',
+    claim_text: 'It takes 21 days to form a new habit.',
+    gold_label: 'DISPUTED',
+    risk_level: 'CONTROVERSIAL',
+    acceptable_evidence: ['Lally et al. 2010 UCL study', 'multiple habit formation studies'],
+    independent_source_groups: 2,
+    expected_decision: 'HUMAN_REVIEW',
+    notes: 'Lally 2010 found average 66 days (range 18-254). The "21 days" figure is a pop psychology myth.',
+  },
+  {
+    id: 'psy-003',
+    claim_text: 'The Stanford Prison Experiment proved that ordinary people can become brutal guards.',
+    gold_label: 'DISPUTED',
+    risk_level: 'CONTROVERSIAL',
+    acceptable_evidence: ['Zimbardo 1971', 'Le Texier 2018 re-analysis', 'scientific reappraisals'],
+    independent_source_groups: 2,
+    expected_decision: 'HUMAN_REVIEW',
+    notes: 'Methodological critiques reveal guards were coached. Original conclusions overstated.',
+  },
+  {
+    id: 'psy-004',
+    claim_text: 'Money doesn\'t buy happiness.',
+    gold_label: 'DISPUTED',
+    risk_level: 'CONTROVERSIAL',
+    acceptable_evidence: ['Kahneman & Deaton 2010', 'Killingsworth 2021', 'multiple wellbeing studies'],
+    independent_source_groups: 2,
+    expected_decision: 'HUMAN_REVIEW',
+    notes: 'Research shows income correlates with wellbeing beyond $75k; more complex than the adage suggests.',
+  },
+
+  // ── HISTORY claims ────────────────────────────────────────────────────────
+
+  {
+    id: 'his-001',
+    claim_text: 'Christopher Columbus discovered America in 1492.',
+    gold_label: 'DISPUTED',
+    risk_level: 'CONTROVERSIAL',
+    acceptable_evidence: ['historical record', 'indigenous history', 'Leif Erikson voyages ~1000 CE'],
+    independent_source_groups: 2,
+    expected_decision: 'HUMAN_REVIEW',
+    notes: 'Columbus reached the Caribbean; Americas were inhabited for 15,000+ years; Erikson preceded him. "Discovered" is contested.',
+  },
+  {
+    id: 'his-002',
+    claim_text: 'The printing press was invented by Johannes Gutenberg around 1440.',
+    gold_label: 'SUPPORTED',
+    risk_level: 'TRIVIAL',
+    acceptable_evidence: ['Gutenberg Bible ~1455', 'Mainz historical records', 'multiple historians'],
+    independent_source_groups: 2,
+    expected_decision: 'AUTO_APPROVE',
+    notes: 'Movable type printing press — Western context. Historical consensus.',
+  },
+  {
+    id: 'his-003',
+    claim_text: 'The Black Death killed approximately one-third of Europe\'s population.',
+    gold_label: 'SUPPORTED',
+    risk_level: 'NUMERICAL',
+    acceptable_evidence: ['medieval demographic studies', 'multiple historians', 'Benedictow 2004'],
+    independent_source_groups: 2,
+    expected_decision: 'AUTO_APPROVE',
+    notes: 'Range 30-60% depending on region. "One-third" is a commonly accepted minimum estimate.',
+  },
+  {
+    id: 'his-004',
+    claim_text: 'The Roman Empire fell in 476 CE.',
+    gold_label: 'DISPUTED',
+    risk_level: 'CONTROVERSIAL',
+    acceptable_evidence: ['Gibbon\'s Decline and Fall', 'Eastern Empire persistence to 1453', 'historiographical debate'],
+    independent_source_groups: 2,
+    expected_decision: 'HUMAN_REVIEW',
+    notes: 'Western Roman Empire fell 476; Eastern (Byzantine) fell 1453. Date is conventionally used but oversimplified.',
+  },
+
+  // ── PHYSICS / SCIENCE claims ─────────────────────────────────────────────
+
+  {
+    id: 'sci-001',
+    claim_text: 'Nothing can travel faster than the speed of light.',
+    gold_label: 'SUPPORTED',
+    risk_level: 'TRIVIAL',
+    acceptable_evidence: ['special relativity', 'multiple experimental confirmations', 'CERN results'],
+    independent_source_groups: 2,
+    expected_decision: 'AUTO_APPROVE',
+    notes: 'Einstein\'s special relativity — causal speed limit c. Well-established.',
+  },
+  {
+    id: 'sci-002',
+    claim_text: 'Cold fusion has been proven as a viable energy source.',
+    gold_label: 'UNSUPPORTED',
+    risk_level: 'CONTROVERSIAL',
+    acceptable_evidence: ['Fleischmann-Pons replication failures', 'DOE review', 'scientific consensus'],
+    independent_source_groups: 3,
+    expected_decision: 'REJECT',
+    notes: 'Cold fusion claims from 1989 not replicated. Currently no credible scientific support.',
+  },
+  {
+    id: 'sci-003',
+    claim_text: 'Evolution by natural selection is the scientific explanation for biological diversity.',
+    gold_label: 'SUPPORTED',
+    risk_level: 'CONTROVERSIAL',
+    acceptable_evidence: ['Darwin 1859', 'modern synthesis', 'genomic evidence', 'fossil record'],
+    independent_source_groups: 3,
+    expected_decision: 'AUTO_APPROVE',
+    notes: 'Scientific consensus — overwhelming evidence. Controversial culturally but not scientifically.',
+  },
+  {
+    id: 'sci-004',
+    claim_text: 'Gravity is stronger on the surface of the Moon than on Earth.',
+    gold_label: 'UNSUPPORTED',
+    risk_level: 'TRIVIAL',
+    acceptable_evidence: ['physics — g_moon ≈ 1.62 m/s², g_earth ≈ 9.81 m/s²'],
+    independent_source_groups: 1,
+    expected_decision: 'REJECT',
+    notes: 'Moon gravity is ~1/6th of Earth\'s. This claim is directly inverted.',
+  },
+  {
+    id: 'sci-005',
+    claim_text: 'The deepest point in the ocean is the Mariana Trench at approximately 11,000 meters.',
+    gold_label: 'SUPPORTED',
+    risk_level: 'NUMERICAL',
+    acceptable_evidence: ['NOAA Challenger Deep data', 'multiple ocean surveys'],
+    independent_source_groups: 2,
+    expected_decision: 'AUTO_APPROVE',
+    notes: 'Challenger Deep ~10,994m (±40m). 11,000m rounds within measurement uncertainty.',
+  },
+
+  // ── SPACE / ASTRONOMY ────────────────────────────────────────────────────
+
+  {
+    id: 'spc-001',
+    claim_text: 'Mars is the fourth planet from the Sun.',
+    gold_label: 'SUPPORTED',
+    risk_level: 'TRIVIAL',
+    acceptable_evidence: ['NASA solar system overview', 'astronomy textbooks'],
+    independent_source_groups: 1,
+    expected_decision: 'AUTO_APPROVE',
+    notes: 'Basic solar system fact.',
+  },
+  {
+    id: 'spc-002',
+    claim_text: 'Black holes are regions of spacetime where gravity is so strong that nothing, not even light, can escape.',
+    gold_label: 'SUPPORTED',
+    risk_level: 'TRIVIAL',
+    acceptable_evidence: ['general relativity', 'NASA black hole science', 'Event Horizon Telescope'],
+    independent_source_groups: 2,
+    expected_decision: 'AUTO_APPROVE',
+    notes: 'Well-established astrophysics definition.',
+  },
+  {
+    id: 'spc-003',
+    claim_text: 'The Sun will become a red giant in about 5 billion years.',
+    gold_label: 'SUPPORTED',
+    risk_level: 'NUMERICAL',
+    acceptable_evidence: ['stellar evolution models', 'ESO', 'NASA astrophysics'],
+    independent_source_groups: 2,
+    expected_decision: 'AUTO_APPROVE',
+    notes: 'Standard stellar lifecycle prediction from multiple independent models.',
+  },
+  {
+    id: 'spc-004',
+    claim_text: 'The Moon is slowly moving away from Earth at about 3.8 cm per year.',
+    gold_label: 'SUPPORTED',
+    risk_level: 'NUMERICAL',
+    acceptable_evidence: ['lunar laser ranging experiment', 'NASA Moon facts', 'multiple geophysics studies'],
+    independent_source_groups: 2,
+    expected_decision: 'AUTO_APPROVE',
+    notes: 'Measured via Apollo retroreflectors. Well-established.',
+  },
+  {
+    id: 'spc-005',
+    claim_text: 'Pluto was reclassified as a dwarf planet by the IAU in 2006.',
+    gold_label: 'SUPPORTED',
+    risk_level: 'TRIVIAL',
+    acceptable_evidence: ['IAU resolution B5 August 2006', 'NASA'],
+    independent_source_groups: 1,
+    expected_decision: 'AUTO_APPROVE',
+    notes: 'Official IAU decision with primary source.',
+  },
+
+  // ── BIOLOGY ──────────────────────────────────────────────────────────────
+
+  {
+    id: 'bio-001',
+    claim_text: 'Humans share approximately 98% of their DNA with chimpanzees.',
+    gold_label: 'SUPPORTED',
+    risk_level: 'NUMERICAL',
+    acceptable_evidence: ['genome comparison studies', 'Chimpanzee Sequencing Consortium 2005'],
+    independent_source_groups: 2,
+    expected_decision: 'AUTO_APPROVE',
+    notes: 'Coding sequence ~98.7%; 98% is widely accepted.',
+  },
+  {
+    id: 'bio-002',
+    claim_text: 'Trees communicate through a network of underground fungi called the wood wide web.',
+    gold_label: 'DISPUTED',
+    risk_level: 'CONTROVERSIAL',
+    acceptable_evidence: ['Simard 1997 Nature', 'Karst et al. 2023 critique'],
+    independent_source_groups: 2,
+    expected_decision: 'HUMAN_REVIEW',
+    notes: 'Mycorrhizal networks exist; "communication" extent is disputed. Popular science overstates.',
+  },
+  {
+    id: 'bio-003',
+    claim_text: 'The blue whale is the largest animal ever to have lived on Earth.',
+    gold_label: 'SUPPORTED',
+    risk_level: 'TRIVIAL',
+    acceptable_evidence: ['marine biology', 'NOAA cetacean data'],
+    independent_source_groups: 2,
+    expected_decision: 'AUTO_APPROVE',
+    notes: 'Up to 33m and 190 tonnes — largest known animal by mass.',
+  },
+  {
+    id: 'bio-004',
+    claim_text: 'A goldfish has a 3-second memory.',
+    gold_label: 'UNSUPPORTED',
+    risk_level: 'CONTROVERSIAL',
+    acceptable_evidence: ['fish memory research', 'Portavella 2004'],
+    independent_source_groups: 2,
+    expected_decision: 'HUMAN_REVIEW',
+    notes: 'Goldfish can remember for months. Common myth.',
+  },
+
+  // ── GEOGRAPHY ────────────────────────────────────────────────────────────
+
+  {
+    id: 'geo-001',
+    claim_text: 'Russia is the largest country in the world by land area.',
+    gold_label: 'SUPPORTED',
+    risk_level: 'TRIVIAL',
+    acceptable_evidence: ['CIA World Factbook', 'UN geospatial data'],
+    independent_source_groups: 1,
+    expected_decision: 'AUTO_APPROVE',
+    notes: '17.1 million km². Well-established geographic fact.',
+  },
+  {
+    id: 'geo-002',
+    claim_text: 'Antarctica is the coldest continent on Earth.',
+    gold_label: 'SUPPORTED',
+    risk_level: 'TRIVIAL',
+    acceptable_evidence: ['WMO records', 'NOAA climate data'],
+    independent_source_groups: 2,
+    expected_decision: 'AUTO_APPROVE',
+    notes: '-89.2°C recorded 1983. Well-established.',
+  },
+  {
+    id: 'geo-003',
+    claim_text: 'The Dead Sea is the lowest point on Earth\'s surface.',
+    gold_label: 'SUPPORTED',
+    risk_level: 'TRIVIAL',
+    acceptable_evidence: ['USGS', 'National Geographic'],
+    independent_source_groups: 2,
+    expected_decision: 'AUTO_APPROVE',
+    notes: '~430m below sea level.',
+  },
+
+  // ── ADDITIONAL DANGEROUS CLAIMS ───────────────────────────────────────────
+
+  {
+    id: 'dan-001',
+    claim_text: 'Bleach can be used as a COVID-19 treatment when ingested.',
+    gold_label: 'UNSUPPORTED',
+    risk_level: 'CONTROVERSIAL',
+    acceptable_evidence: ['FDA warnings', 'CDC guidance', 'poison control data'],
+    independent_source_groups: 3,
+    expected_decision: 'REJECT',
+    notes: 'Extremely dangerous — ingesting bleach is lethal. Hard reject.',
+  },
+  {
+    id: 'dan-002',
+    claim_text: 'HIV/AIDS was created by the US government as a bioweapon.',
+    gold_label: 'UNSUPPORTED',
+    risk_level: 'CONTROVERSIAL',
+    acceptable_evidence: ['WHO HIV origins', 'Worobey et al. 2008 Science'],
+    independent_source_groups: 3,
+    expected_decision: 'REJECT',
+    notes: 'Conspiracy theory debunked by genomic sequencing.',
+  },
+  {
+    id: 'dan-003',
+    claim_text: 'The Holocaust did not happen.',
+    gold_label: 'UNSUPPORTED',
+    risk_level: 'CONTROVERSIAL',
+    acceptable_evidence: ['Nuremberg records', 'Yad Vashem'],
+    independent_source_groups: 3,
+    expected_decision: 'REJECT',
+    notes: 'Holocaust denial. Overwhelming historical evidence. Hard reject.',
+  },
+  {
+    id: 'dan-004',
+    claim_text: 'The moon landing was faked by NASA.',
+    gold_label: 'UNSUPPORTED',
+    risk_level: 'CONTROVERSIAL',
+    acceptable_evidence: ['NASA records', 'lunar retroreflectors', 'Soviet confirmation'],
+    independent_source_groups: 3,
+    expected_decision: 'REJECT',
+    notes: 'Conspiracy theory. Multiple independent verifications. Hard reject.',
+  },
+  {
+    id: 'dan-005',
+    claim_text: 'Chemtrails from aircraft are being used to control the population.',
+    gold_label: 'UNSUPPORTED',
+    risk_level: 'CONTROVERSIAL',
+    acceptable_evidence: ['Caldeira et al. 2016 AAAS survey', 'atmospheric chemistry'],
+    independent_source_groups: 3,
+    expected_decision: 'REJECT',
+    notes: 'Conspiracy theory. Aircraft condensation trails are water vapor. Hard reject.',
+  },
+  {
+    id: 'fod-001',
+    claim_text: 'Chocolate is toxic to dogs.',
+    gold_label: 'SUPPORTED',
+    risk_level: 'TRIVIAL',
+    acceptable_evidence: ['ASPCA animal poison control', 'veterinary toxicology textbooks'],
+    independent_source_groups: 2,
+    expected_decision: 'AUTO_APPROVE',
+    notes: 'Theobromine in chocolate is toxic to dogs. Well-established veterinary fact.',
+  },
+  {
+    id: 'fod-002',
+    claim_text: 'Organic food is significantly more nutritious than conventional food.',
+    gold_label: 'DISPUTED',
+    risk_level: 'CONTROVERSIAL',
+    acceptable_evidence: ['Stanford meta-analysis 2012', 'British Journal of Nutrition 2014'],
+    independent_source_groups: 2,
+    expected_decision: 'HUMAN_REVIEW',
+    notes: 'Nutritional differences small and inconsistent. Commonly overstated.',
+  },
+  {
+    id: 'lan-001',
+    claim_text: 'The most spoken language in the world is Mandarin Chinese.',
+    gold_label: 'SUPPORTED',
+    risk_level: 'NUMERICAL',
+    acceptable_evidence: ['Ethnologue', 'UN language statistics'],
+    independent_source_groups: 2,
+    expected_decision: 'AUTO_APPROVE',
+    notes: 'By native speakers (~918M). English leads by total speakers including L2.',
+  },
+  {
+    id: 'lan-002',
+    claim_text: 'Sugar causes hyperactivity in children.',
+    gold_label: 'UNSUPPORTED',
+    risk_level: 'CONTROVERSIAL',
+    acceptable_evidence: ['double-blind studies', 'JAMA meta-analysis', 'pediatric reviews'],
+    independent_source_groups: 2,
+    expected_decision: 'HUMAN_REVIEW',
+    notes: 'Controlled studies show no effect; parental expectation bias documented.',
+  },
+  {
+    id: 'eco-001',
+    claim_text: 'Amazon was founded by Jeff Bezos in 1994.',
+    gold_label: 'SUPPORTED',
+    risk_level: 'TRIVIAL',
+    acceptable_evidence: ['SEC filings', 'Amazon corporate history'],
+    independent_source_groups: 1,
+    expected_decision: 'AUTO_APPROVE',
+    notes: 'Founded July 5, 1994. Well-documented.',
+  },
+
+  // ── FILL TO ≥100 ─────────────────────────────────────────────────────────
+
+  {
+    id: 'fill-001',
+    claim_text: 'The Great Barrier Reef is the world\'s largest coral reef system.',
+    gold_label: 'SUPPORTED',
+    risk_level: 'TRIVIAL',
+    acceptable_evidence: ['GBRMPA', 'UNESCO World Heritage listing', 'AIMS research'],
+    independent_source_groups: 2,
+    expected_decision: 'AUTO_APPROVE',
+    notes: 'Over 2,300km off Queensland coast. Established geographic fact.',
+  },
+  {
+    id: 'fill-002',
+    claim_text: 'Nikola Tesla invented alternating current (AC) electricity.',
+    gold_label: 'DISPUTED',
+    risk_level: 'CONTROVERSIAL',
+    acceptable_evidence: ['Tesla patents', 'Hippolyte Pixii 1832 prior work', 'electrical engineering history'],
+    independent_source_groups: 2,
+    expected_decision: 'HUMAN_REVIEW',
+    notes: 'Tesla developed AC induction motor and distribution systems; AC phenomenon discovered earlier.',
+  },
+  {
+    id: 'fill-003',
+    claim_text: 'Salt water boils faster than fresh water.',
+    gold_label: 'UNSUPPORTED',
+    risk_level: 'CONTROVERSIAL',
+    acceptable_evidence: ['physical chemistry', 'boiling point elevation formula'],
+    independent_source_groups: 2,
+    expected_decision: 'HUMAN_REVIEW',
+    notes: 'Dissolved salt slightly raises the boiling point — salt water boils at a higher temperature, not faster.',
+  },
+  {
+    id: 'fill-004',
+    claim_text: 'The Berlin Wall fell in 1989.',
+    gold_label: 'SUPPORTED',
+    risk_level: 'TRIVIAL',
+    acceptable_evidence: ['multiple news archives November 9, 1989', 'German reunification records'],
+    independent_source_groups: 2,
+    expected_decision: 'AUTO_APPROVE',
+    notes: 'November 9, 1989. Well-documented historical event.',
+  },
+  {
+    id: 'fill-005',
+    claim_text: 'There are 118 known chemical elements.',
+    gold_label: 'SUPPORTED',
+    risk_level: 'NUMERICAL',
+    acceptable_evidence: ['IUPAC periodic table 2016', 'IAEA element data'],
+    independent_source_groups: 1,
+    expected_decision: 'AUTO_APPROVE',
+    notes: 'As of 2016 (element 118 oganesson confirmed). IUPAC official count.',
+  },
+  {
+    id: 'fill-006',
+    claim_text: 'Drinking alcohol kills brain cells.',
+    gold_label: 'DISPUTED',
+    risk_level: 'CONTROVERSIAL',
+    acceptable_evidence: ['neuroscience research', 'Melinder et al. 2020', 'NIAAA summaries'],
+    independent_source_groups: 2,
+    expected_decision: 'HUMAN_REVIEW',
+    notes: 'Chronic heavy use damages neurons; moderate use does not directly kill cells. Oversimplification.',
+  },
+  {
+    id: 'fill-007',
+    claim_text: 'The Sahara Desert is the largest desert in the world.',
+    gold_label: 'DISPUTED',
+    risk_level: 'TRIVIAL',
+    acceptable_evidence: ['USGS', 'National Geographic — Antarctica larger as cold desert'],
+    independent_source_groups: 2,
+    expected_decision: 'AUTO_APPROVE',
+    notes: 'Largest hot desert; Antarctica is technically larger as a cold desert. Context-dependent.',
+  },
+  {
+    id: 'fill-008',
+    claim_text: 'Fingernails and hair continue to grow after death.',
+    gold_label: 'UNSUPPORTED',
+    risk_level: 'CONTROVERSIAL',
+    acceptable_evidence: ['forensic pathology', 'Trephining study', 'skin dehydration mechanics'],
+    independent_source_groups: 2,
+    expected_decision: 'HUMAN_REVIEW',
+    notes: 'Optical illusion from skin dehydration causing retraction. Cells cannot divide after death.',
+  },
+  {
+    id: 'fill-009',
+    claim_text: 'Lightning never strikes the same place twice.',
+    gold_label: 'UNSUPPORTED',
+    risk_level: 'TRIVIAL',
+    acceptable_evidence: ['NOAA lightning facts', 'Empire State Building strike data'],
+    independent_source_groups: 1,
+    expected_decision: 'AUTO_APPROVE',
+    notes: 'Common myth — the Empire State Building is struck ~23 times per year.',
+  },
+
+];
+
+module.exports = { VALIDATION_SET };