when will stockfish improvement slow down • page 1/1 • General Chess Discussion • lichess.org

stockfish 19 gained 40 elo from the already very strong stockfish 18 , when will that slow down and stockfish getting close to being max optimized ?

Kitsune9Tails

Well you're getting better engine quicker than ever. But if you don't like it, well, unlucky for you it will never stop improving...

Ender88

@najdorfplayer48 said ^

stockfish 19 gained 40 elo from the already very strong stockfish 18 , when will that slow down and stockfish getting close to being max optimized ?

I think when the game will be solved you will see no engine improvement

@najdorfplayer48 said [^](/forum/redirect/post/AhYcrgk1) > stockfish 19 gained 40 elo from the already very strong stockfish 18 , when will that slow down and stockfish getting close to being max optimized ? I think when the game will be solved you will see no engine improvement

najdorfplayer48

edited

@Ender88 said ^

stockfish 19 gained 40 elo from the already very strong stockfish 18 , when will that slow down and stockfish getting close to being max optimized ?

I think when the game will be solved you will see no engine improvement

that would be far in the future if not impossible , i meant when will the neural network , pvs will be close to max optimized

@Ender88 said [^](/forum/redirect/post/zJ8TiSsR) > > stockfish 19 gained 40 elo from the already very strong stockfish 18 , when will that slow down and stockfish getting close to being max optimized ? > > I think when the game will be solved you will see no engine improvement that would be far in the future if not impossible , i meant when will the neural network , pvs will be close to max optimized

lizani

Fred Friedel, co-founder of Chessbase, believes engines will plateau around 3300.

KMcGeoch

I'm unsure if ELO is best measurement for engines for the simple reason that there is inflation and/or deflation in ratings. So if you take a 2500 from now, 10 years ago and 30 years ago they're likely to have quite different real strengths.

So you have the question that if engines will plateau at 3300 ELO then what will 3300 ELO plateau at?

Although I would imagine what OP meant is Stockfish 19 is performing against Stockfish 18 with score that would be expected for a 40 ELO difference. As far as improvements go usually it follows Pareto principle where 80% of the gains are done with 20% of the work. So once you get the low lying fruit you get smaller gains that require a lot more work to obtain.

I'm unsure if ELO is best measurement for engines for the simple reason that there is inflation and/or deflation in ratings. So if you take a 2500 from now, 10 years ago and 30 years ago they're likely to have quite different real strengths. So you have the question that if engines will plateau at 3300 ELO then what will 3300 ELO plateau at? Although I would imagine what OP meant is Stockfish 19 is performing against Stockfish 18 with score that would be expected for a 40 ELO difference. As far as improvements go usually it follows Pareto principle where 80% of the gains are done with 20% of the work. So once you get the low lying fruit you get smaller gains that require a lot more work to obtain.