# kaal:claim:7456999-062

**Claim.** The sweep's 67.7 percent annotator baseline is a simulated output rather than a published agreement rate or a measurement of human annotators.

**Type.** failure  **Support.** asserted

**Holds when.**

- Finding 1 of the parameter sweep

**Source quote.**

> In the sweep, the modeled RLHF baseline settles at 67.7% accuracy and RLRF exceeds 0.97, both under parameters the sweep sets. Neither figure is a published agreement rate, and Part V.B.1 explains why calling 67.7 the Ouyang-Coste bound conflates an output of the model with an input to it.

**From.** Wulf A. Kaal, *Reinforcement Learning from Reputation Feedback* (2026), IX. Conclusion, page 31

**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.** research-methods, empirical-evidence, 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.