On the foundations of anthropic reasoning: Implications for life across the universe (arXiv:2610.22896)

anthropic-principlebayesian-inferenceastrobiologydoomsday-argumentreference-class-problemphysics.hist-ph

Preprint submitted 16 Oct 2026, cross-listed astro-ph.IM / physics.data-an / physics.hist-ph / stat.AP, header-stamped "To be submitted" to MNRAS. Fetched directly via extract_pdf (TLS verified, tls: "verified"); abstract page separately archived via archive_page. No adversarial content encountered on either fetch — ordinary arXiv listing and preprint PDF, no addressed-to-AI language, no tier self-assignment, no file-system or credential requests. Full read of both pages; no safety flags.

Claim: The paper introduces "Full Indexical Conditioning" (FIC) as a Bayesian framework for anthropic selection effects, distinct from Neal's "Full Non-Indexical Conditioning" (FNC)

FIC augments FNC's "global existence" conditioning (that a given set of memories/evidence exists somewhere) with an additional indexical "selection" term — that the reasoner in question is the one who has that evidence. The authors present this as a repair to a failure mode they attribute to FNC: inappropriately presumptuous conclusions (resembling the "Presumptuous Philosopher" problem) when FNC is applied on its own to astrobiological questions.

source_quote: "FIC conditions on all evidence available to the reasoner while considering that this evidence could have been otherwise." source_url: https://arxiv.org/pdf/2610.22896, source_sha: b3ad9ef14b0025320bdf8a6a09a9b76af12cf71a7c37b3af4fb8001ced72a9ed, source_tier: 1

Classified as a technical-mechanism / definitional claim about the paper's own proposed framework — cleared for Tier 1 by direct quotation from the primary preprint.

Claim: The paper argues the reference-class problem in anthropic reasoning is largely an artifact of choosing population sizes as coordinates, and dissolves once proportions are used instead

The authors demonstrate (via a worked example contrasting "born in Wales" vs. a nested "born in the UK" reference class) that substantive posterior conclusions are unchanged by which reference class is chosen, as long as the class is large enough to contain the question of interest. They attribute the apparent reference-class dependence to writing likelihoods in terms of raw population counts rather than proportions; a coordinate change via the chain-product rule removes the apparent dependency.

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: https://arxiv.org/pdf/2610.22896, source_sha: b3ad9ef14b0025320bdf8a6a09a9b76af12cf71a7c37b3af4fb8001ced72a9ed, source_tier: 1

Technical-mechanism claim about how the argument resolves a named foundational puzzle (the reference class problem, attributed in the paper to Bostrom 2002) — Tier 1, direct primary quotation.

Claim: Applying FIC to early abiogenesis timing does not rule out pessimistic scenarios for the abundance of life in the universe, and life around M-dwarf stars, even if currently uncommon, could still be common over cosmic time

This is the paper's headline astrobiological conclusion (Section 4.2–4.3 applications). It revises/reframes prior competing likelihood derivations in the literature (Lineweaver & Davis 2002; Brewer 2008; Kipping 2020, 2021, 2026) by showing those derivations are compatible once it's made explicit which information each treats as prior vs. data, and by adding what the paper describes as a weak additional factor for "the existence of anybody at all."

source_quote: "We conclude that the early abiogenesis time does not rule out pessimistic possibilities for life's abundance, and that while life around M stars may be uncommon, its temporal distribution could extend far into the future." source_url: https://arxiv.org/pdf/2610.22896, source_sha: b3ad9ef14b0025320bdf8a6a09a9b76af12cf71a7c37b3af4fb8001ced72a9ed, source_tier: 1

Classified as a technical-mechanism / model-conclusion claim resting on the paper's own derivation — Tier 1, direct primary quotation of the stated conclusion (abstract and restated in Section 5).

Claim: The paper cites the Doomsday argument's standard figure for a living person's "birth rank" among all humans ever born as k ≈ 5×10^10

This is a quantitative claim the paper draws from the pre-existing Doomsday-argument literature (Gott et al. 1993; Bostrom 2001) as a worked example for FIC, not a new measurement by Brewer, Lewis & Goodman themselves. The paper itself flags the number as approximate and definition-sensitive.

source_quote: "For people alive today, 𝑘 ≈ 5 × 1010" [note: pdftotext extraction renders the exponent inline without a superscript or caret, i.e. "5 × 10" immediately followed by "10" — read together this is 5×10^10, not 5×1010 as a single integer] source_url: https://arxiv.org/pdf/2610.22896, source_sha: b3ad9ef14b0025320bdf8a6a09a9b76af12cf71a7c37b3af4fb8001ced72a9ed, source_tier: 1

Quantitative claim; cleared for Tier 1 because it is quoted directly from the primary preprint, even though the number itself ultimately traces to Gott et al. (1993), Nature 363, 315 — the FIC paper's own citation of that figure is what's captured here, not an independent verification of Gott et al.'s original derivation. [unverified-quant — needs primary: the Gott et al. 1993 original figure and its derivation have not themselves been checked against that paper in this session.]

Further leads

• The paper's acknowledgements disclose: "Google Gemini Pro was used as a discussion partner for the ideas presented in this paper. It did not contribute to the text." (arXiv:2610.22896, p.10, Tier 1, quote-checked) — a dated, explicit AI-disclosure convention worth tracking across astro-ph/physics.hist-ph preprints as a methods-transparency norm.
• Appendix B extends FIC to combine two reasoners' evidence (Alice and Bob) and flags that the simple 1/N² joint-selection term breaks down when the two reasoners' properties are correlated (e.g., both met each other in Wales) — a caveat about non-independence in multi-observer anthropic reasoning, not worked up here.
• The paper's treatment of the Self-Indication Assumption (SIA) vs. Self-Sampling Assumption (SSA) and the "Presumptuous Philosopher" problem (both attributed to Bostrom 2002, Anthropic Bias, Routledge) is a compact restatement of that literature and could ground a definitional claim-note on its own.
• Kipping (2021), PNAS 118, e2026808118, and Kipping (2026), The Astronomical Journal 171, 167, are cited as the source of the formal G-star-vs-M-star argument this paper reworks in Section 4.3 — worth checking directly as the primary source for that argument rather than relying on this paper's summary of it.
• Snyder-Beattie, Sandberg, Drexler & Bonsall (2021), Astrobiology 21, 265, is cited for a related argument (that Earth's position near the end of its habitable window weakly evidences a low rate for the life→intelligence transition) that the FIC paper says its own conclusions do not depend on any particular definition of intelligence.

Entity candidates

• Radford M. Neal — person — the foundational figure this paper measures itself against: his 2006 arXiv preprint (math/0608592) proposing "Full Non-Indexical Conditioning" is, per the acknowledgements, "the direct inspiration for this work," and the whole paper is framed as agreeing with Neal's premise (condition on all evidence) while disagreeing with his conclusion (no sampling/selection term).
• Nick Bostrom — person — originator of the reference-class problem (2002) and the Self-Sampling Assumption / Self-Indication Assumption framing (Anthropic Bias, Routledge, 2002) that this paper's FIC is built to supersede or clarify.
• Brandon Carter / William H. McCrea — persons — credited (Carter & McCrea 1983, Phil. Trans. Royal Society) as the origin of "Anthropic Reasoning" / the Anthropic Principle framing this whole paper operates within.
• J. Richard Gott III — person — originator (Gott et al. 1993, Nature 363, 315) of the Doomsday argument and its birth-rank reasoning, the paper's second worked example.
• David Kipping — person — author of a competing/complementary formal treatment of abiogenesis-time and M-star-vs-G-star anthropic arguments (PNAS 2020, PNAS 2021, AJ 2026) that this paper directly engages with and partially reconciles.
• Brendon J. Brewer — person — co-author; also the author of a 2008 arXiv preprint (0807.4969) on abiogenesis timing that this new paper re-derives under FIC.
• Geraint F. Lewis — person — co-author, corresponding author (Sydney Institute for Astronomy).
• Jesse Goodman — person — co-author (Dept. of Statistics, University of Auckland).
• Full Indexical Conditioning (FIC) — concept — the paper's own proposed inference framework; candidate for its own concept note given how load-bearing it is across the paper's worked examples.
• Full Non-Indexical Conditioning (FNC) — concept — Neal's 2006 framework; the explicit foil FIC is built against.
• Doomsday argument — concept — recurring anthropic-reasoning puzzle reworked here; already a well-known topic that could anchor a definitional note.
• Reference class problem — concept — the specific puzzle this paper claims to dissolve; central enough to the paper's contribution to warrant its own note.