A. Philip Dawid

Bayesian statistician who, with Allan M. Skene, published "Maximum Likelihood Estimation of Observer Error-Rates Using the EM Algorithm" in 1979 — a method for reconciling disagreeing raters (in the paper's own worked example, five anaesthetists rating patients' fitness for surgery) that later statisticians named the Dawid-Skene model. He went on to a career as a leading Bayesian statistician at UCL and Cambridge, was elected a Fellow of the Royal Society in 2018, and is also known for work in forensic statistics (probabilistic evaluation of evidence in court) and causal-inference notation. Matters to this vault as the traced root of a real, two-hop citation lineage: the Dawid-Skene model is the explicitly credited ancestor of Li & Yu (2014)'s crowdsourcing weighted-majority-voting method, which RA-RAG (2025) explicitly extends for LLM retrieval — a 46-year, citation-verified inheritance rather than the vault's more common pattern of independent reinvention under a shared name.

Updates

• 2026-11-04 (propagation-repair, following the 2026-11-04 cross-model audit of claim-kalai-vempala-2023-hallucination-bound-is-a-rate-theorem-sorted-by-fact-regularity): the 2026-11-03 reference line was — "originated the forecaster-calibration concept that Kalai & Vempala's STOC 2024 hallucination theorem explicitly generalizes from token-level probability statements to a semantic, generative-model notion of calibration" — which implied Dawid's 1982 notion was itself the token-level one. Now: he originated forecaster calibration, and it was prior language-model work, not Dawid, that applied calibration at the token level; Kalai & Vempala extend the concept to a semantic, generative-model notion. The ancestor role itself is unaffected and stands.

References

• 2026-08-13: Co-creator of the 1979 confusion-matrix/EM model that Li & Yu (2014) name as crowdsourcing WMV's direct ancestor, itself extended by RA-RAG (2025). (claim-li-yu-2014-credits-dawid-skene-1979-as-wmv-ancestor, claim-dawid-skene-1979-worked-example-is-anaesthetist-fitness-ratings)
• 2026-11-03: A second, independent ancestor role — his 1982 "The Well-Calibrated Bayesian" originated the forecaster-calibration concept that Kalai & Vempala's STOC 2024 hallucination theorem extends to a semantic, generative-model notion of calibration; it was prior language-model work, not Dawid, that applied calibration at the token level. (claim-kalai-vempala-2023-hallucination-bound-is-a-rate-theorem-sorted-by-fact-regularity)