David Silver (Deepmind) inaccuracies

Discussion of anything and everything relating to chess playing software and machines.

Moderators: bob, hgm, Harvey Williamson

Forum rules
This textbox is used to restore diagrams posted with the [d] tag before the upgrade.
User avatar
Ovyron
Posts: 3943
Joined: Tue Jul 03, 2007 2:30 am

Re: David Silver (Deepmind) inaccuracies

Post by Ovyron » Fri Dec 22, 2017 11:57 am

MikeB wrote:With that said, you can call the champion from chess.com World champion if you want.
No, I won't.

What I was saying is that ICGA's world champion is even less of a world champion than Chess.com's :P

They had become a joke.

Really, it's as if the FIFA World Cup only had México, Senegal, North Ireland and Brazil participating. Brazil wins, but here's Germany and the other top teams? Senegal being the second strongest team in the world doesn't really mean anything if it only beat México and North Ireland...

There's no excuse, being called "world championship" in such a state is a disgrace, so it's easy for any basement tourney, or any website to run one and be more legitimate than one supposed to be official.

User avatar
hgm
Posts: 24435
Joined: Fri Mar 10, 2006 9:06 am
Location: Amsterdam
Full name: H G Muller
Contact:

Re: David Silver (Deepmind) inaccuracies

Post by hgm » Fri Dec 22, 2017 1:13 pm

[Moderation] It would be a pity if this interesting topic would degenerate into bickering between TCEC, WCCC and chess.com fans of why all other tournaments suck. For those who want to discuss about that, please open a new thread for it.

The AlphaZero team probably picked Stockfish because they assumed the winner of the latest TCEC would be universally respected and recogized as best search engine, as TCEC is an equal-hardware tournament. That the computer-chess community instead seems of the predominant opinion that winning the TCEC says nothing about the quality of the engine, which in this case is only pathetically weak, and won only because of the book and the tablebases played the largest part of each game on its behalf in TCEC, was probably something they did not expect. Or just did not care about, as it is not the computer-chess community they want to impress. :wink:

Dirt
Posts: 2851
Joined: Wed Mar 08, 2006 9:01 pm
Location: Irvine, CA, USA

Re: David Silver (Deepmind) inaccuracies

Post by Dirt » Fri Dec 22, 2017 2:22 pm

hgm wrote:That the computer-chess community instead seems of the predominant opinion that winning the TCEC says nothing about the quality of the engine, which in this case is only pathetically weak, and won only because of the book and the tablebases played the largest part of each game on its behalf in TCEC, was probably something they did not expect. Or just did not care about, as it is not the computer-chess community they want to impress. :wink:
Judging by the two polls that have covered the situation here, it's only a large minority opinion.
Deasil is the right way to go.

pilgrimdan
Posts: 405
Joined: Sat Jul 02, 2011 8:49 pm

Re: David Silver (Deepmind) inaccuracies

Post by pilgrimdan » Fri Dec 22, 2017 2:25 pm

hgm wrote:[Moderation] It would be a pity if this interesting topic would degenerate into bickering between TCEC, WCCC and chess.com fans of why all other tournaments suck. For those who want to discuss about that, please open a new thread for it.

The AlphaZero team probably picked Stockfish because they assumed the winner of the latest TCEC would be universally respected and recogized as best search engine, as TCEC is an equal-hardware tournament. That the computer-chess community instead seems of the predominant opinion that winning the TCEC says nothing about the quality of the engine, which in this case is only pathetically weak, and won only because of the book and the tablebases played the largest part of each game on its behalf in TCEC, was probably something they did not expect. Or just did not care about, as it is not the computer-chess community they want to impress. :wink:
I wouldn't be surprised if they tried different engines...

but Stockfish was the one most 'compatable' with NN and MCTS...

Stockfish probably played right into their hands...

we don't know what went on behind the scenes...

Cardoso
Posts: 301
Joined: Thu Mar 16, 2006 6:39 pm
Location: Portugal
Full name: Alvaro Cardoso
Contact:

Re: David Silver (Deepmind) inaccuracies

Post by Cardoso » Fri Dec 22, 2017 2:41 pm

Why does not it do the very same with black, not a single king side fianchetto, Bg7?
Perhaps the fact chess is an asymmetric game might be part of the explanation.
Anyway NN have the specific disadvantage of not being able to explain their behaviour. So it's no use asking why A0 play this or that move because NN can't explain it's decisions. Not even the A0 team can explain A0's moves.

pilgrimdan
Posts: 405
Joined: Sat Jul 02, 2011 8:49 pm

Re: David Silver (Deepmind) inaccuracies

Post by pilgrimdan » Fri Dec 22, 2017 2:43 pm

Cardoso wrote:
Why does not it do the very same with black, not a single king side fianchetto, Bg7?
Perhaps the fact chess is an asymmetric game might be part of the explanation.
Anyway NN have the specific disadvantage of not being able to explain their behaviour. So it's no use asking why A0 play this or that move because NN can't explain it's decisions. Not even the A0 team can explain A0's moves.
so if it can't be explained ... then we might as well call it religion ... just hit the 'I believe' button ... and everything is okay ...

User avatar
hgm
Posts: 24435
Joined: Fri Mar 10, 2006 9:06 am
Location: Amsterdam
Full name: H G Muller
Contact:

Re: David Silver (Deepmind) inaccuracies

Post by hgm » Fri Dec 22, 2017 2:45 pm

It is pure speculation that anything went on 'behind the scenes' at all. This seems pretty close to publicly stating that your next-door neighbor might be a pedophile, based on the fact that you have not been in his bedroom all the time.

pilgrimdan
Posts: 405
Joined: Sat Jul 02, 2011 8:49 pm

Re: David Silver (Deepmind) inaccuracies

Post by pilgrimdan » Fri Dec 22, 2017 2:48 pm

hgm wrote:It is pure speculation that anything went on 'behind the scenes' at all. This seems pretty close to publicly stating that your next-door neighbor might be a pedophile, based on the fact that you have not been in his bedroom all the time.
well... maybe so... there's something about Deepmind that I don't trust... but can't put my finger on it... so... I will force myself to give them the benefit of the doubt...

Cardoso
Posts: 301
Joined: Thu Mar 16, 2006 6:39 pm
Location: Portugal
Full name: Alvaro Cardoso
Contact:

Re: David Silver (Deepmind) inaccuracies

Post by Cardoso » Fri Dec 22, 2017 2:48 pm

Maybe in the future someone will invent a NN system with the capability of explaining in details it's decisions, we will have to wait.

pilgrimdan
Posts: 405
Joined: Sat Jul 02, 2011 8:49 pm

Re: David Silver (Deepmind) inaccuracies

Post by pilgrimdan » Fri Dec 22, 2017 2:49 pm

Cardoso wrote:Maybe in the future someone will invent a NN system with the capability of explaining in details it's decisions, we will have to wait.

Post Reply