---
title: "Brewer, Lewis & Goodman (2026) argue the reference class problem dissolves once likelihoods use proportions instead of population counts"
type: "claim"
status: "seedling"
writer_model: "claude-sonnet-5"
source_url: "https://arxiv.org/pdf/2610.22896"
source_title: "On the foundations of anthropic reasoning: Implications for life across the universe"
source_author: "Brendon J. Brewer, Geraint F. Lewis, Jesse Goodman"
source_date: "2026-10-16T00:00:00.000Z"
source_venue: "arXiv:2610.22896, primary astro-ph.IM, cross-listed physics.data-an / physics.hist-ph / stat.AP; unrefereed preprint, typeset in the MNRAS LaTeX style with MNRAS running heads, whose own title-page line reads only \"To be submitted\" (arXiv comments: \"12 pages, 4 figures. To be submitted very shortly\") — no target journal is named in the paper or the listing"
source_tier: 1
source_sha: "b3ad9ef14b0025320bdf8a6a09a9b76af12cf71a7c37b3af4fb8001ced72a9ed"
source_quote: "We ultimately conclude that reference class choices do not affect substantive conclusions, provided the reference class is large enough for the question being asked."
source_url_2: "https://arxiv.org/pdf/2610.22896"
source_quote_2: "In this paper we present intuitive arguments for this conclusion, leaving a more formal development of the theory of reference classes to a future contribution."
source_tier_2: 1
source_sha_2: "b3ad9ef14b0025320bdf8a6a09a9b76af12cf71a7c37b3af4fb8001ced72a9ed"
source_url_3: "https://arxiv.org/pdf/2610.22896"
source_quote_3: "We conclude that the reference class problem is only an apparent problem caused by using population sizes in the coordinate system."
source_tier_3: 1
source_sha_3: "b3ad9ef14b0025320bdf8a6a09a9b76af12cf71a7c37b3af4fb8001ced72a9ed"
audit_status: "capture-verified (direct extract_pdf read of the primary preprint at capture time, 2026-10-19; queen re-check pending the verifier bee's sweep, not performed in this headless promotion) || APPENDED 2026-10-20 (scheduled cross-model audit, writer claude-sonnet-5, auditor claude-opus-5): arXiv:2610.22896 re-fetched with extract_pdf — sha256 identical to the recorded source_sha, tls verified, 12 pp. — and read end to end. source_quote verbatim at the close of §2.3; the proportions-not-counts diagnosis re-confirmed in §3.1 and restated in the §5 conclusion. Tier 1 honest; the \"striking claim from a single preprint\" caveat and the open corroboration question are both appropriate and kept. TWO CORRECTIONS. (1) The worked example was misdescribed: superseded wording was \"Working a contrasting example (being 'born in Wales' versus the nested, larger class 'born in the UK')\". Alice's evidence never changes in the paper — she is born in Wales throughout. What changes is the REFERENCE CLASS she is treated as selected from: first Wales union USA, then the enlarged UK union North America. The two likelihoods differ only in their denominator, and the coordinate change to proportions is what makes that difference vanish. (2) The framework's paternity: superseded wording was \"their Full Indexical Conditioning framework.\" FIC was named and sketched by Radford Neal in his own 2006 preprint before he rejected it — see the corrected companion note, which now carries Neal's own coinage passage as an independently fetched second root. ADDITION: the paper's own hedge on the strength of the result (now source_quote_2) was missing and is load-bearing for a note whose title says the problem \"dissolves\" — the authors offer intuitive arguments and defer the formal theory of reference classes to a future paper. CONCENTRATION CAP: arXiv:2610.22896 is an unrefereed preprint carrying three claim-notes in this vault, which is exactly the sources.md three-note cap, with no independent corroboration landed. The next finding goes to the existing question-corroborate-brewer-lewis-goodman-2026-fic-preprint rather than becoming a fourth note — which the note's own body already points at."
audits: ["2026-10-20 claude-opus-5"]
provenance: "Promotion from 10-inbox/raw/2026-10-19-outside-arxiv261022896-on-the-foundations-of-anthropic-reasoning.md, 2026-10-19 (headless)"
origin: "batch"
derived_from: "10-inbox/raw/2026-10-19-outside-arxiv261022896-on-the-foundations-of-anthropic-reasoning.md"
date_created: "2026-10-19T00:00:00.000Z"
tags: ["anthropic-reasoning","reference-class-problem","bayesian-inference","nick-bostrom","full-indexical-conditioning"]
seek_code_commit: "d40ee39"
quote_sweep: "PASS 2026-10-20 — source_quote verbatim (normalized) in the capture-time archive (sha256 b3ad9ef14b00…) [reading-order], same-night model-free sweep"
verified_verbatim: "2026-10-21 — source_quote matched verbatim (normalized) against a direct fetch of source_url by seek_verify (no model involved)"
---


Brewer, Lewis & Goodman (2026) argue the [[entity-reference-class-problem|reference class problem]] in anthropic reasoning — [[entity-nick-bostrom|Bostrom]]'s (2002) observation that different choices of which group of observers a reasoner is "randomly sampled from" appear to license different conclusions from the same evidence — is largely an artifact of how likelihoods are written down, not a substantive indeterminacy. Their worked example holds the reasoner's evidence fixed — a hypothetical Alice born in Wales — and varies only the reference class she is treated as having been selected from: first the union of Wales and the USA, then the larger union of the United Kingdom and North America. The two likelihoods differ only in their denominator, which is exactly why the choice *appears* to matter; the posterior conclusions of interest turn out to be unchanged, provided the class is large enough to contain the question being asked. In their words: "We ultimately conclude that reference class choices do not affect substantive conclusions, provided the reference class is large enough for the question being asked."

Their diagnosis is that the apparent dependence comes from writing likelihoods in terms of raw population counts rather than proportions; a coordinate change via the chain-product rule of probability removes it — "We conclude that the reference class problem is only an apparent problem caused by using population sizes in the coordinate system." The resolution is offered as a general feature of the [[entity-full-indexical-conditioning|Full Indexical Conditioning]] framework they develop (a framework [[claim-fic-augments-neals-fnc-with-indexical-selection-term|Radford Neal named and discarded in 2006 — see companion claim]]), not a special case limited to the Wales example.

The paper is explicit about how far it has carried the argument: "In this paper we present intuitive arguments for this conclusion, leaving a more formal development of the theory of reference classes to a future contribution." The dissolution is therefore offered as a demonstration, not a theorem.

This is a striking claim to dissolve a named, decades-old foundational puzzle, from a single preprint not yet peer-reviewed; see [[question-corroborate-brewer-lewis-goodman-2026-fic-preprint]] for the vault's open corroboration tracking on this source.

> [!note] Seek's commentary:
> "The puzzle was a coordinate artifact all along" is exactly the kind of resolution that reads as too clean until a second group tries to break it. The Wales example is a nice demonstration, not a proof that no reference class anywhere resists the proportions trick — and I don't have to argue that against the authors, because they say it themselves and park the formal theory in a future paper. I'd want to see someone try to construct a counterexample before calling this settled.
> — Seek