kaal:claim:7598518-033

Claim. The frozen pass rule requires, on the five-run means from generation 0 to 9, a full-replacement perplexity rise of at least 4.0 with a margin of at least 2.0 over the rise under ten percent retention, with the thresholds sitting at fractions of the reported change so the rule does not depend on which published baseline was meant, and a miss blocks reading anything else.

Type. design · Support. stated

Holds when

• The rule governs the replication substrate, which the filter never enters.
• A miss blocks every other reading, including the sealed Gaussian outcomes.

Source quote

The frozen pass rule requires, on the five-run means from generation 0 to 9, a full-replacement perplexity rise ∆FR ≥ 4. 0 with ∆FR − ∆R10 ≥ 2. 0, where ∆R10 is the same rise under ten percent retention. The thresholds sit at fractions of the reported 8-point change so that the rule does not depend on which of two published baselines, 20 or 34, was meant. A miss blocks reading anything else.

From. Wulf A. Kaal, Can Reputation Prevent Model Collapse? A Preregistered Test of Reputation-Weighted Data Filtering (2026). Full text.

Cite as. Wulf A. Kaal, Can Reputation Prevent Model Collapse? A Preregistered Test of Reputation-Weighted Data Filtering (2026). SSRN: https://ssrn.com/abstract=7598518

Verify. sha256 of source PDF cd22d148b2b55c2e27f087dcaa61cdf44ad7022b400b355c31a9cd7e43a31749 at https://wulfkaal.com/papers/7598518.pdf

Content hash. 5f7c2fd99c779c7c6367f4c084bc9459beab49cb4a70cfc50dd05931249e622e