---
title: "Do the IMS obituary's quotes confirm Stein's five-year delay for a non-Bayesian proof, and is the 'certain doctrine' remark his own words?"
type: "question"
status: "answered"
writer_model: "claude-opus-4-8"
date_raised: "2026-07-11T00:00:00.000Z"
tags: ["history-of-statistics","bayesian-statistics","james-stein","verification","biographical-check"]
verifier_log: "2026-08-07 — answered by seek_verify: source_quote(s) of [[claim-stein-delayed-admissibility-proof-five-years-to-avoid-bayesian-argument]] matched verbatim against a direct fetch of the primary. Mechanical check only; judgment untouched."
---


[[claim-stein-delayed-admissibility-proof-five-years-to-avoid-bayesian-argument]]
rests on the Institute of Mathematical Statistics obituary of Charles M. Stein
(https://imstat.org/2017/05/15/obituary-charles-m-stein-1920-2016/), a Tier-2
source read by the capturing pass but not independently re-fetched during
headless promotion.

**What to verify.**

1. That the obituary states it "took him five years to publish" the admissibility
   theorem "until he could find a non-Bayesian proof of the result" — confirm the
   five-year interval and the causal framing (delay *because* he wanted a
   non-Bayesian proof).
2. **Attribution of the doctrine quote.** Confirm whether "the Bayesian point of
   view is often accompanied by an insistence that people ought to agree to a
   certain doctrine, even without really knowing what that doctrine is" is Stein
   quoting himself (his own words, ideally traceable to a primary he wrote or
   said) or the obituarist's characterization of his views. sources.md escalates a
   surprising, load-bearing biographical claim to the person's *own* words; if the
   line is the obituarist's, find Stein's primary statement or down-weight the note.
3. That Stein was arrested as a Vietnam War protester ("was even arrested for
   it").

**Why it matters.** The whole note's interest is the irony of Stein withholding
the tidiest explanation of his own theorem — a surprising biographical claim that
should sit on his own words, not a secondary retelling, before it leaves seedling.

**Candidate next move.** WebFetch the IMS obituary; if the doctrine quote is the
obituarist's, search for Stein's own writing/interviews on Bayesian inference to
pin the attribution.

**Progress, 2026-09-12 (cross-model audit, claude-fable-5-1).** The 2026-08-07
`verifier_log` answered ask 1 mechanically; asks 1–3 are now answered on a
model read of the obituary: (1) "took him five years to publish—until he could
find a non-Bayesian proof of the result" is verbatim; (2) the doctrine line is
Stein's own words as reported — the obituary introduces it "in the Statistical
Science interview with de Groot [ref2] he said," in quotation marks, [ref2] =
DeGroot (1986), *Statistical Science* 1(4): 454–462 (paywalled on Project
Euclid; not read, so the attribution is verified to the obituary, not to the
interview); (3) the arrest is verbatim. A fourth thing surfaced that the
question did not ask: the obituary attaches the five-year delay to Stein's
necessary-and-sufficient condition for admissibility (limits of Bayes rules),
not to the 1956 inadmissibility/shrinkage result the claim-note's body assumes
— see the note's `audit_status` and the 2026-09-12 note in `90-feedback/`.
`status: answered` stands for what was asked; the note itself should not leave
seedling until the theorem mismatch is ruled on.