Page 2 of 2

Re: 19/08/2019 New Poll

Posted: Mon Jan 13, 2020 3:20 am
by attackgm
I think it will be some sort of AI nn engine they will become highly advanced.

Re: 19/08/2019 New Poll

Posted: Thu Apr 02, 2020 8:32 pm
by chessplayer
If that is true, do you think hand tuned opening books will become irrelevant?

Re: 19/08/2019 New Poll

Posted: Mon Apr 06, 2020 7:49 pm
by KhalidOmar
chessplayer wrote: Thu Apr 02, 2020 8:32 pm If that is true, do you think hand tuned opening books will become irrelevant?
I don't think so, even if NN or A/B engines become much stronger. The human touch is still unique and powerful in ways engines will never understand. That is why hand tuned books will still play a significant role in chess!

Re: 19/08/2019 New Poll

Posted: Mon May 18, 2020 4:32 pm
by Gustl
Well, human touch may work differently today as compared to 10 years ago. Noone can soundly say "I'll keep that variation in book even though a 3600 elo-engine marks it as bad". However, the human touch may be to complement LC0's lines with an A/B engine or vice versa in intuitive ways, considering specifc input depending on varying parameters or such.

Re: 19/08/2019 New Poll

Posted: Mon May 25, 2020 9:35 am
by kutchess
DavidG wrote: Fri Aug 30, 2019 5:26 pm Gentlemen,
I believe that in 10 years, computer chess will have advanced to the stage where every game will be a draw. Lines will be so perfect, and blunders so rare, such as to render the game of chess "solved". This will increase the popularity of ancillary chess games such as Fischer Random chess.
David
I do not agree. In 10 years it is perhaps known that one or two opening moves (1.e4 e5 or 1.d4 Nf6) are best and that all such games will be a draw. And furthermore that all other opening moves will badly loose. :)
swissman

Re: 19/08/2019 New Poll

Posted: Tue Jun 02, 2020 7:38 am
by Russell3162001
Until a Quantum machine appears, it is likely that a hybrid NN-AB engine is likely to achieve dominance in the medium term with a more pure NN engine coming to fore in the longer run

Re: 19/08/2019 New Poll

Posted: Tue Jun 02, 2020 7:53 am
by SimonMilos
I don't think the tactical skill of AB engines will be beaten with Neural Networks.
The "easy", but quite convincing answer is that an AB engine developed around an NN's eval will be best.

Re: 19/08/2019 New Poll

Posted: Tue Jun 02, 2020 3:44 pm
by Russell3162001
The question the becomes if the NN Evaluation is compatible with an A-B search or is it the combination of the NN Eval and MCTS search that drives the success of NN engines