# Review: priority credit across alternative validation routes ## 1. Precise restatement with assumptions made explicit **Setup.** A discovery is a monotone "completion" structure on a set of papers. Its minimal sufficient sets ("routes") are: - R₁ = {a, c} - R₂ = {b, d₁, …, dₘ}, with m ≥ 1. **Assumptions:** - **(A1) Both routes are minimal.** No proper subset of either route completes the discovery. With m ≥ 1, this means no single paper is sufficient. - **(A2) Everything eventually appears.** All m+3 papers are published, and the publication order is uniformly random over all (m+3)! permutations. Equivalently, arrival times are i.i.d. continuous. This is the hidden assumption that defines "expected credit." - **(A3) Winner-take-all priority.** All credit goes to the single paper whose appearance first makes some route complete. **Claim.** Papers a and b have the same value under two measures: - standalone sufficiency (both zero); - indispensability to the full archive (both zero, since removing either one leaves the other route complete). Yet their expected credits differ by the ratio (m+1)²/4. This ratio grows quadratically in the length of the longer route. ## 2. Checking the steps and re-deriving independently **Standalone sufficiency and indispensability.** - Standalone: neither {a} nor {b} is sufficient when m ≥ 1. Correct. - Indispensability: removing a leaves R₂ intact, and removing b leaves R₁ intact. Correct. **Paper a.** Let a arrive at time t. - Paper a completes R₁ at t if and only if c < t, which has probability t. - Paper a is the *first* completion if and only if R₂ has not completed by t. That requires max(b, d's) > t, which has probability 1 − t^{m+1}. - These two events involve disjoint sets of papers, so they are independent. P(a) = ∫₀¹ t(1 − t^{m+1}) dt = 1/2 − 1/(m+3) = (m+1)/(2(m+3)). ✓ **Paper b.** Let b arrive at time t. - Paper b completes R₂ at t if all d's arrived before t, which has probability t^m. - Route R₁ must not be complete yet, which has probability 1 − t². P(b) = ∫₀¹ (t^m − t^{m+2}) dt = 1/(m+1) − 1/(m+3) = 2/((m+1)(m+3)). ✓ **Ratio.** [(m+1)/(2(m+3))] · [(m+1)(m+3)/2] = (m+1)²/4. ✓ **Independent combinatorial cross-check.** - R₁ wins if and only if the last paper overall lies in R₂. That has probability (m+1)/(m+3). - Within the winning route, each member is equally likely to be the completer, by exchangeability. - So P(a) = (m+1)/(2(m+3)) and P(b) = (2/(m+3))/(m+1). Both match the integrals. - The probabilities sum to 1: 2·(m+1)/(2(m+3)) + (m+1)·2/((m+1)(m+3)) = 1. **Worked numbers.** | m | Total papers | P(a), P(c) | P(b), each P(dᵢ) | Ratio | |---|---|---|---|---| | 1 | 4 | 1/4 | 1/4 | 1 (symmetric case) | | 2 | 5 | 3/10 | 2/15 | 9/4 | For m = 2, the total is 0.6 + 0.4 = 1. ## 3. Attempts to break the claim **m = 0.** Then b alone is sufficient, which violates A1. The construction needs m ≥ 1, and the text's premises exclude m = 0. No problem. **"Quadratically."** - The ratio is about m²/4. - The absolute gap is not quadratic: P(a) tends to 1/2, while P(b) is about 2/m². - So the phrase is accurate only as a statement about the ratio, read as a function of route length. Credit dispersion grows quadratically. **Robustness to the credit rule.** Suppose credit is shared equally among all papers of the winning route instead of going only to the completer. The values are identical: P(a) is half of (m+1)/(m+3), and P(b) is 1/(m+1) of 2/(m+3). So the result does not depend on the winner-take-all detail. **Dependence on A2.** The result does depend on uniform random order. If R₂'s papers tend to arrive earlier, the ratio can shrink or reverse. The claim says only that the papers "*can* differ," and the text explicitly fixes equal publication-speed distributions. So this is a stated modelling choice, not a flaw. **Conclusion of this section.** I found no counterexample. The existential claim is fully proved by the construction. ## 4. Novelty Under a uniformly random order, the "first completer" credit is exactly the **Shapley–Shubik power index** of the simple game whose minimal winning coalitions are R₁ and R₂. The paper a is pivotal precisely when its predecessors form a losing coalition that becomes winning once a is added. It is standard in cooperative game theory that the following do not determine a player's power index: - singleton worth (here, zero); - marginal contribution to the grand coalition (here, zero, the "indispensability" measure). Players in shorter minimal winning coalitions get more power. The computation is therefore a textbook-style Shapley–Shubik example. The framing in terms of priority and scientific credit is a nice application, but the mathematics is a special case of known results. ## 5. Fairness of the one-sentence summary The sentence accurately reflects the text: - It is existential ("can differ"). - It names the credit rule. - The two measures on which a and b agree match what the text defines. Minor omissions: - It does not mention the uniform-random-order assumption, though "expected" implicitly needs a probability model and the text states it. - "Quadratically" refers to the credit ratio as route length grows, not to the absolute difference. Neither omission is a material overstatement. ```json {"verdict": "correct", "novelty": "special case or variant of known results", "fair_summary": "yes", "confidence": 5, "one_line": "The integrals and the (m+1)²/4 ratio check out exactly. The result is an instance of Shapley–Shubik power indices differing among players with equal standalone and grand-coalition marginal value, under the stated uniform-random-order model."} ```