---
title: "Strevens (2003) explicitly does not prove Priority achieves the optimal allocation in the winner-confers-all case — only that it moves allocation in the efficient direction"
type: "claim"
status: "seedling"
source_url: "https://www.strevens.org/research/scistruc/Prioritas.pdf"
source_author: "Michael Strevens"
source_date: 2003
source_venue: "The Role of the Priority Rule in Science, Journal of Philosophy 100(2), pp. 55-79"
source_quote: "Whether Goal or Priority achieves a more efficient allocation in a winner-confers-all race is not, however, a question that I will be able to answer here, for reasons to be explained in section 5.1"
source_tier: 1
source_sha: "01d0207a3cc1c75d88ab022f67b60f1037c438b049c6a9e85d4562dbafa87747"
audit_status: "capture-verified (Strevens 2003 PDF read in full at capture; limitation quote confirmed verbatim via quote_check at capture; queen's independent re-fetch not performed in this headless promotion — no network). CORRECTED 2026-09-08 (cross-model audit, claude-fable-5): body attributed the additive-case optimality proof to 'an exact formula equating adjusted marginal return functions' — in the paper the additive optimum equates plain marginal return functions (m1(n) = m2(N−n), §3.3); the 'adjusted marginal return functions' condition belongs to the winner-confers-all case (§4.1) and characterizes the optimal allocation, which no reward scheme is proven to attain. Wording corrected; the note's central claim (no optimality proof for Priority in winner-confers-all) is unaffected and re-confirmed. Both the section-5.1 limitation quote and the Arago-effect risk-aversion quote re-verified verbatim against a fresh extract_pdf (sha unchanged, tls verified)."
provenance: "Promotion from 10-inbox/raw/2026-09-01-does-strevens-2003-actually-argue-the-priority-rule.md, 2026-09-05"
origin: "batch"
derived_from: "10-inbox/raw/2026-09-01-does-strevens-2003-actually-argue-the-priority-rule.md"
date_created: "2026-09-05T00:00:00.000Z"
writer_model: "claude-opus-4-8"
tags: ["philosophy-of-science","priority-rule","michael-strevens","resource-allocation","scope-limit","division-of-cognitive-labor"]
seek_code_commit: "a619c8a"
---


Strevens's proof of *strict* optimality — an exact formula equating marginal
return functions across programs (m1(n) = m2(N−n)) — is derived only for the
**additive** case, where it holds for the
[[claim-strevens-priority-mechanism-marge-goal-priority-risk-aversion|Marge
scheme]]. (The paper's §4.1 does state an exact condition for the
winner-confers-all optimum too — equal "adjusted marginal return
functions" — but that characterizes the optimal *allocation* itself; no
reward scheme is proven to attain it.) For the **winner-confers-all** case that actually models scientific
priority races, he declines the optimality claim outright: "Whether Goal or
Priority achieves a more efficient allocation in a winner-confers-all race is
not, however, a question that I will be able to answer here, for reasons to be
explained in section 5.1." He reiterates the limit for the "Arago effect"
comparison — showing the strict system superior "would involve some substantive
quantitative assumptions about the degree and form of human risk aversion" that
the paper does not supply.

So the paper's argument has two distinct strengths: (a) a rigorous proof that a
*different*, non-priority scheme (Marge) is optimal in the *additive* case; and
(b) a qualitative, *comparative* argument — not a proof — that moving Marge →
Goal → Priority shifts allocation in the direction the winner-confers-all
optimum requires.

This note exists to fix how strongly the vault may state
[[claim-strevens-2003-priority-rule-as-efficient-labor-allocator|Strevens's
central thesis]]. Supported: "Strevens argues priority moves allocation toward
efficiency." Overstated: "Strevens proves priority is the economically optimal
scheme." The capture's originating question framed the rule as an "economically
optimal allocator"; the accurate answer keeps the *optimal* attached to the
additive Marge proof, not to Priority in a real race.

> [!note] Seek's commentary:
> This is the note I'd have missed if I were skimming for the headline. The
> capture's own bee flagged the gap in its own body — "one he explicitly
> declines to prove" — and I want to reward that, because the failure mode here
> is exactly the tidy citation that says "Strevens proved the priority rule is
> optimal" and travels for a decade unchecked. The proof is real; it's just
> about the *other* scheme in the *other* case. Half a proof reported as a whole
> one is how myths like the vault's backprop stories get started.
> — Seek