Page 5 of 5
Re: Lc0 question
Posted: Wed Jul 10, 2019 1:52 am
by Ovyron
mwyoung wrote: ↑Wed Jul 10, 2019 1:44 am
Ovyron wrote: ↑Wed Jul 10, 2019 1:30 am
mclane wrote: ↑Sun Jul 07, 2019 9:44 pmIsn’t it enough Stockfish and Komodo and Houdini do this race ??
What race? Stockfish won that one by a mile long time ago. If I recall correctly Komodo topped the rating lists only once, and for a very short while, so I'm not sure Stockfish was in danger of losing that race at any point.
The statement is irrational. Why would would want to limit progress on a game that is impossible to be solved.
No, it is not enough for Stockfish, Komodo, and Houdini to do this race.
New programs, and fresh ideas are always welcome.
Lc0, and other NN engines are the fresh ideas at this moment. And showing Stockfish has not won the race....
In that case, all chess machines are in the race, and with big improvements, some "dark horse" engine could appear and top the lists, even above Leela, even without NNs. So it makes no sense to only mention the top three A/B engines.
Maybe someone at home is cooking up the next Rybka 1.0 beta, of our times, when Alpha Zero appeared, I thought it was it, but now that Leela is stronger than Alpha, its advantage over second best is nowhere near the biggest advantage of Rybka over second best. Some chess entity could have 3600 ELO, and could beat both Stockfish and Leela by a big margin, all we need is an ingenious idea, one that isn't just implemented in Stockfish to shoot it up...
Re: Lc0 question
Posted: Wed Jul 10, 2019 3:03 am
by mwyoung
Ovyron wrote: ↑Wed Jul 10, 2019 1:52 am
mwyoung wrote: ↑Wed Jul 10, 2019 1:44 am
Ovyron wrote: ↑Wed Jul 10, 2019 1:30 am
mclane wrote: ↑Sun Jul 07, 2019 9:44 pmIsn’t it enough Stockfish and Komodo and Houdini do this race ??
What race? Stockfish won that one by a mile long time ago. If I recall correctly Komodo topped the rating lists only once, and for a very short while, so I'm not sure Stockfish was in danger of losing that race at any point.
The statement is irrational. Why would would want to limit progress on a game that is impossible to be solved.
No, it is not enough for Stockfish, Komodo, and Houdini to do this race.
New programs, and fresh ideas are always welcome.
Lc0, and other NN engines are the fresh ideas at this moment. And showing Stockfish has not won the race....
In that case, all chess machines are in the race, and with big improvements, some "dark horse" engine could appear and top the lists, even above Leela, even without NNs. So it makes no sense to only mention the top three A/B engines.
Maybe someone at home is cooking up the next Rybka 1.0 beta, of our times, when Alpha Zero appeared, I thought it was it, but now that Leela is stronger than Alpha, its advantage over second best is nowhere near the biggest advantage of Rybka over second best. Some chess entity could have 3600 ELO, and could beat both Stockfish and Leela by a big margin, all we need is an ingenious idea, one that isn't just implemented in Stockfish to shoot it up...
I think that is what I said. And why his statement is irrational. New programs, and fresh ideas are always welcome. The last time I checked no one heard of Lc0 two years ago, and the best chess engines are stronger this year, then last year. And the best is not Stockfish, Komodo, or Houdini.
Re: Lc0 question
Posted: Wed Jul 10, 2019 6:19 am
by mclane
The idea of LC0 is not to win against stockfish,
The idea of LC0 is to play a better chess. A chess on a different quality level.
Not the stupid machine chess stockfish is doing,
Why would anyone feel the need to do a race with stockfish ??
The competition between LC0 and others is not about elo.
It is about chess understanding and playing chess on a different understanding level.
Re: Lc0 question
Posted: Wed Jul 10, 2019 7:58 am
by Ovyron
mclane wrote: ↑Wed Jul 10, 2019 6:19 am
The idea of LC0 is not to win against stockfish,
The idea of LC0 is to play a better chess. A chess on a different quality level.
Not the stupid machine chess stockfish is doing,
Why would anyone feel the need to do a race with stockfish ??
The competition between LC0 and others is not about elo.
It is about chess understanding and playing chess on a different understanding level.
I don't think your argument holds, it can easily be flipped to this:
The idea of Stockfish is not to win against LC0,
The idea of Stockfish is to play a better chess. A chess on a different quality level.
Not the tactically blind chess LC0 is doing,
Why would anyone feel the need to do a race with LC0 ??
The competition between Stockfish and others is not about elo.
It is about chess understanding and playing chess on a different understanding level.
What's the difference?
Re: Lc0 question
Posted: Wed Jul 10, 2019 9:31 am
by jp
Yep. No engine of any type "understands" chess, so the only way they can help humans understand chess is to produce better chess moves through different types of brute force calculation. But of course then humans need to work out when different engines are producing better chess moves and when they are not. That is not always easy to do.
Re: Lc0 question
Posted: Thu Jul 11, 2019 8:55 pm
by ThatsIt
CEGT 40/4 update is online now:
List:
http://www.cegt.net/blitz.htm
Best single list:
http://www.cegt.net/40_4_Ratinglist/40_ ... liste.html
Best wishes,
G.S.
(CEGT team)