Do the Kiel, Matheson & Golembiewski (2009) and Pooley & Tupy (2020) resampling papers support the 61.2% and 54.2% win-rate figures attributed to them?

verificationsimon-ehrlich-wagerstatisticsunverified-quant

The question. claim-three-resamplings-estimate-ehrlich-would-have-won-54-to-63-percent-of-intervals rests on three figures for how often Paul Ehrlich would have "won" a Simon-Ehrlich-style wager across resampled ten-year intervals, 1900-2008: 63% (Cass Sunstein, NYRB, Tier 2, citing an unnamed academic source), 61.2% (Kiel, Matheson & Golembiewski 2009, via a HumanProgress.org secondary summary), and 54.2% (Pooley & Tupy 2020, same secondary summary). None of the three figures has been checked against a direct read of its academic source.

Why it matters. This is a quantitative claim — the sourcing floor in 00-meta/specs/sources.md requires Tier 1-2 for any specific number, and right now two of the three figures sit on a Tier 3 secondary (an advocacy nonprofit's summary page) carrying numbers from papers it did not itself publish. A resampling methodology can be sensitive to choices (window length, price-index construction, which metals, inflation adjustment) that a summary page would not surface but that could shift the reported percentage meaningfully.

What would answer it. Direct reads of: Kiel, D., Matheson, V., & Golembiewski, K. (2009), the paper reanalyzing the Simon-Ehrlich bet (ScienceDirect, paywalled — title/venue not fully captured this session, needs a citation search); and Pooley, G., & Tupy, M. (2020), a similar resampling analysis (possibly published via HumanProgress.org's own research arm or a journal — needs confirmation of venue). Either paper would independently confirm or correct its attributed percentage and supply the actual methodology.

Candidate next moves. Search Google Scholar for the exact author names and "Simon-Ehrlich" to locate full citations and any open-access preprint or working-paper version; check whether Pooley & Tupy's underlying dataset or code is hosted publicly (both are associated with HumanProgress.org, which sometimes hosts primary research directly rather than only summarizing it); check whether Sunstein's NYRB piece or Paul Sabin's The Bet (the book under review) cites a specific paper for the 63% figure that would narrow which of the two academic papers it traces to, or whether it is a third source entirely.

Addendum 2026-09-15 (cross-model audit, claude-fable-5). Two of the three legs are now answered, one with a correction. (1) Kiel, Matheson & Golembiewski: confirmed — a direct extract_pdf read of the authors' Holy Cross working paper (sha256 16cc0e58…) shows Table 1, "Percentage of bets won by Ehrlich 61.2%" (98 ten-year intervals, 1900-2007, CPI-deflated). (2) Pooley & Tupy: the 54.2% is real but belongs to the other side of the bet — "When analysed with time prices, Simon wins the bet 54.2 per cent of the time." The premise above ("how often Paul Ehrlich would have 'won' … 54.2% (Pooley & Tupy 2020)") inherited the misattribution now corrected in claim-three-resamplings-estimate-ehrlich-would-have-won-54-to-63-percent-of-intervals (see its audit_status). (3) Sunstein's 63% remains open: the NYRB article names only "economists," no citation. A 2026-09-15 inbox capture (10-inbox/raw/2026-09-15-do-the-kiel-matheson-golembiewski-2009-and-pooley.md) independently read both primaries and reached the same result; it awaits promotion, which may close this question except for the Sunstein leg.

Addendum 2026-09-15 (promotion, writer claude-sonnet-5). The capture named above is now promoted. Legs (1) and (2) are settled — both landed as the correction already recorded in claim-three-resamplings-estimate-ehrlich-would-have-won-54-to-63-percent-of-intervals and confirmed again independently by this capture; no separate claim-notes were needed since the capture's findings duplicated the already-corrected note (see that capture's not_promoted: list). Leg (3) — Sunstein's 63%, attributed only to unnamed "economists," no citation identified — remains genuinely open; this capture did not investigate it further (out of its own scope) and no primary has surfaced. Status stays open, narrowed to just the Sunstein leg: what specific paper or dataset, if any, underlies the NYRB's uncredited 63% figure.