Dojo Talks: A Chess Podcast EP 147 | FIDE Ratings & The Stats of Cheating (ft. Dr. Kenneth Regan)
14 snips
Mar 25, 2025 Dr. Kenneth Regan, a computer scientist specializing in chess cheating detection, dives into the fascinating world of FIDE ratings. He reveals how recent changes have led to inflated ratings, particularly for players under 2000. The conversation touches on the challenges of accurately measuring skill and detecting cheating, where statistical analysis plays a crucial role. Regan shares insights on discrepancies faced by players post-pandemic and discusses the critical need for reliable data in both online and over-the-board chess.
AI Snips
Chapters
Transcript
Episode notes
FIDE Rating Boost Corrected Deflation
- The 2024 FIDE rating boost corrected deflation, especially below 2000 rating, rather than inflating ratings arbitrarily.
- Previously, low-rated players showed stronger performance than their ratings indicated, causing deflation at higher levels.
Rating Scale Compression Below 2000
- Ratings between 1000 and 2000 were "spaghettied out," stretching the ELO scale beyond realistic expectations.
- The correct difference between ratings in this range should be about 150 points, not 200, leading to Jeff Sonis' proposal of raising the rating floor to 1400.
Top Players Show Mild Deflation
- High-rated players showed mild deflation over the last decade, losing about 30 to 50 rating points.
- Pandemic disruptions caused juniors to be underrated temporarily, which balanced deflation in older players.