# kaal:claim:7456999-008

**Claim.** Among thirty-six repeatedly measured parties, checked-case records predict withheld-case accuracy at a rank correlation of +0.781, and parties failing checked cases passed withheld cases in zero of 168 observations.

**Type.** empirical  **Support.** evidenced

**Holds when.**

- the measured cohort and parties with repeated measurement

**Source quote.**

> The measured cohort of 564 records, produced by twenty distinct model families against twenty-three speci fications and scored by a deterministic checker on cases withheld from the producing party, establishes that a party’s record where the work is checked predicts its accuracy where it is not, at a rank correlation of +0.781 across thirty-six parties with repeated measurement, and that a party failing the checked cases passes the withheld ones in zero of one hundred and sixty-eight observations.

**From.** Wulf A. Kaal, *Reinforcement Learning from Reputation Feedback* (2026), Abstract, page 1

**Cite as.** Wulf A. Kaal, Reinforcement Learning from Reputation Feedback (2026). SSRN: https://ssrn.com/abstract=7456999

**Verify.** sha256 of source PDF `01016fbbcc25cad205cebd59ea686d8dfee8ae9b632095d86ba6c9a900456a4b` at https://raw.githubusercontent.com/wulfkaal/Academic-Papers/main/papers/pdf/Kaal%20-%202026%20-%20Reinforcement%20Learning%20from%20Reputation%20Feedback.pdf

**Topics.** empirical-evidence, reputation, ai-and-agents

**Canonical form.** This markdown file is the canonical hashed representation of the claim. Its sha256 is the content hash used for attestation.