kaal:claim:7598518-014

Claim. A filter that claims to address collapse must beat real-data retention, not only unfiltered replacement, because retaining ten percent of the original data slows the damage.

Type. condition · Support. argued

Holds when

• The comparator standard follows from published results on real-data retention.
• It sets a requirement on what a filter must show, not a result about any filter.

Source quote

Retaining ten percent of the original data slows the damage (Shumailov et al. 2024), and accumulating all real data keeps test error bounded in the settings studied (Gerstgrasser et al. 2024; Kazdan et al. 2025), although even a synthetic share as small as 1 per 1,000 can stop larger training sets from helping in regression (Dohmatob et al. 2025). A filter that claims to address collapse must therefore beat real-data retention, not only unfiltered replacement.

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. 52afd2495b702fd12ce6bfb580bfd07b9da8b49f2203263bb3ad5127ef66b17d