Chess.com Computer Chess Championship

[Houdini]

UCI config:
32 threads.
4096 GB hash.
6-man Syzygy TBs (HDD).
All other UCI options will be set to default.

Tournament specifics:
Ponder off.
No opening book.
No draw adjudication.
No endgame adjudication. (The endgames be played out to mate or a forced draw for the benefit of the viewers.)

I guess "4096 GB hash" is a typo, should be "4096 MB"?

It's a pity that all games will be played from the starting position, that means that a lot of interesting openings (e.g. Sicilian or King's Indian) will never be played. We'll probably see mostly French and Queen's Gambit declined.

[kranium]

UCI config:
32 threads.
4096 GB hash.
6-man Syzygy TBs (HDD).
All other UCI options will be set to default.

Tournament specifics:
Ponder off.
No opening book.
No draw adjudication.
No endgame adjudication. (The endgames be played out to mate or a forced draw for the benefit of the viewers.)

I guess "4096 GB hash" is a typo, should be "4096 MB"?

It's a pity that all games will be played from the starting position, that means that a lot of interesting openings (e.g. Sicilian or King's Indian) will never be played. We'll probably see mostly French and Queen's Gambit declined.

whoops, yes, should read 4 GB
thanks for clarifying that Robert

In my experience, tournaments with 'no opening book' do in fact provide a great variety of openings.
I have run several of them in the last few years and posted detailed results and stats here on talkchess.
http://talkchess.com/forum/viewtopic.php?t=55795

Here is a link to the opening statistics summary from one of them:
http://talkchess.com/forum/viewtopic.ph ... 145#619145
You can also find detailed individual engine stats earlier in the topic.

as you can see the King's Indian and Sicilian were also played, along with many others.

IMO, The variety demonstrated is more than enough for a small rapid/blitz tournament like CCCC...
and it will be fascinating to see what moves/lines these top engines will chose.

[Houdini]

Here is a link to the result and opening stats from one of them:
http://talkchess.com/forum/viewtopic.ph ... 145#619145

as you can see the King's Indian and Sicilian were also played.

Interesting, indeed the diversity is larger than I imagined.
But from the 280 games only 2 are King's Indian and only 2 Grünfeld -
compared to 52 French, and 76 Queen's Gambit Declined/Semi-Slav.

Even in this small tournament people may get bored after a couple of dozen games to see yet another French/QGD.
Already a 2-move opening book could create a lot more diversity and still provide a level playing field for all engines.

[JJJ]

Yeah I agree. Shortest book possible but more opening.

[hgm]

As Uri correctly suggests, 50-move and 3-fold draw rules will certainly be in effect...

Better correct the rules then. Because that is an adjudication. Otherwise participants that end up in a cursed win could rightfully protest against being adjudicated as 50-move draw.

[kranium]

Here is a link to the result and opening stats from one of them:
http://talkchess.com/forum/viewtopic.ph ... 145#619145

as you can see the King's Indian and Sicilian were also played.

Interesting, indeed the diversity is larger than I imagined.
But from the 280 games only 2 are King's Indian and only 2 Grünfeld -
compared to 52 French, and 76 Queen's Gambit Declined/Semi-Slav.

Even in this small tournament people may get bored after a couple of dozen games to see yet another French/QGD.
Already a 2-move opening book could create a lot more diversity and still provide a level playing field for all engines.

Thanks Robert (and Jean) for the feedback.

Don't forget, in the preliminary rapid rounds, each engine plays each other engine only twice...
and with 10 different engines, 90 games total...I'm sure we'll see some real variety, plenty of Spanish, etc.

The superfinal is a different story however, with 2 strongest engines playing 20 games (or more if tiebreak is necessary) against each other.
I agree: a 2 move book would be a good idea here I think. I will consider it and mention it to the team.

However, the chess.com team feels strongly about letting the engines make their own moves, without human interference.
It may be a hard sell...

[kranium]

As Uri correctly suggests, 50-move and 3-fold draw rules will certainly be in effect...

Better correct the rules then. Because that is an adjudication. Otherwise participants that end up in a cursed win could rightfully protest against being adjudicated as 50-move draw.

Yes absolutely...good point, don't want any confusion on that end.
Thanks much Harm

[hgm]

To be fully compliant with FIDE rules, you could of course adjudicate only after 70 reversible moves, but accept engine claims already after 50.

[Dirt]

To be fully compliant with FIDE rules, you could of course adjudicate only after 70 reversible moves, but accept engine claims already after 50. :wink:

And isn't it a six fold repetition for a draw? But only an idiot or FIDE would follow that rule. (Hmm, I think I'm being redundant there.)

[hgm]

Yes, but for repetitions it doesn't make much of a difference, because forced wins never need to repeat at all. Forced wins can run aground on the 50-move rule, however.