Should we have fervent faith in computer analysis? • page 3/3 • General Chess Discussion • lichess.org

Xenobite edited

Rank Engine Participations Games W D L W (%) D (%) L (%) Pts Avg Pts Finals Trophies
1 Stockfish 15 1771 484 1161 126 27.33 65.56 7.11 1064.5 0.601 14 8
2 Houdini 15 1264 316 805 143 25 63.69 11.31 718.5 0.568 6 4
3 Komodo 15 1412 366 908 138 25.92 64.31 9.77 820 0.581 7 3
4 LCZero 4 452 107 292 53 23.67 64.60 11.73 253 0.560 4 2

One question, is LCzero an update of alphazero, if so does anyone know what makes it different? I also notice here in the table linked, the LCzero bot played 4 participations, the smallest there, yet won 107 games while losing 53. I am not that well adept to the rules, why did it participate less? Rank Engine Participations Games W D L W (%) D (%) L (%) Pts Avg Pts Finals Trophies 1 Stockfish 15 1771 484 1161 126 27.33 65.56 7.11 1064.5 0.601 14 8 2 Houdini 15 1264 316 805 143 25 63.69 11.31 718.5 0.568 6 4 3 Komodo 15 1412 366 908 138 25.92 64.31 9.77 820 0.581 7 3 4 LCZero 4 452 107 292 53 23.67 64.60 11.73 253 0.560 4 2

playerx3

First, if you are using engine analysis for improvement, it is recommended use it after you do manual analysis.
Second, engine used it makes almost no difference compared to hardware resources you give it. For example, lichess analysis marked move as blunder, which turned to actually be best move, if you let engine run little longer.

First, if you are using engine analysis for improvement, it is recommended use it after you do manual analysis. Second, engine used it makes almost no difference compared to hardware resources you give it. For example, lichess analysis marked move as blunder, which turned to actually be best move, if you let engine run little longer.