The TCEC...

Discussion of computer chess matches and engine tournaments.

Moderators: bob, hgm, Harvey Williamson

Forum rules
This textbox is used to restore diagrams posted with the [d] tag before the upgrade.
Lyudmil Tsvetkov
Posts: 6052
Joined: Tue Jun 12, 2012 10:41 am

Re: The TCEC...

Post by Lyudmil Tsvetkov » Sat Nov 26, 2016 11:35 am

Uri Blass wrote:
Carotino wrote:Adam, your mama did not teach you education? Why do you insult me?
Well, we fly over the peasants and talking about chess.

This is the variant used in the last game:

1.e4 e5 2Nc3 Nf6 3.Bc4 Nxe4 4.Qh5 Nd6 5.Bb3 Nc6? 6.Nb5! g6 7.Qf3 f5? 8.Qd5 Qf6 9.Nxc7+ Kd8 10.Nxa8 b6 11.Qf3 Bb7 12.h4 Bh6? 13.d3 f4?

Of course, Black (Houdini) lost the game.What sense does this thing?
Stockfish did not lose playing black with the same opening against Houdini.

I guess that the target of the openings is to help the stronger program to win.
With balanced opening there is a danger that we get 100 draws(and even if it is only 95 draws then it will be a disappointment for the spectators) so using unbalanced position may be the only way to see which engine is stronger because engines are almost unbeatable in balanced opening and I guess in some years we are going to have 100% draws in long time control games between top engines unless you choose opening like that opening when it is not clear if white win or if it is a draw.
any 300-elo-stronger engine than current tops will score something like +65 =32 -3 in TCEC, no matter the opening lines.

we just got too accustomed to very equal competition (elo within 20-30 points) in last couple of eiditions.

look at early editions, the winning rate is much higher.

beram
Posts: 1187
Joined: Wed Jan 06, 2010 2:11 pm

Re: The TCEC...

Post by beram » Sat Nov 26, 2016 4:01 pm

Modern Times wrote:Time control is way too long for me to even think about watching any games. Increasing the number of cores was a good opportunity to shorten the time control in my opinion. However you can look at the games afterwards if that is an issue I suppose.

What they have done with TCEC is incredibly impressive though, all credit to those involved.
I agree with you Ray and with Graham, that thx must go out to those involved with TCEC
On the other hand I share your criticism on the way too long time control

The opening choice i find interesting and varied enough, but the treshold for first move out of book could imo have been set slightly lower e.g. <0,6

To Ljudmil I wonna say be realistic, we have these two amazingly strong engines and we don't have any 300 ELO stronger other engine for now or within 6-8 years

Carotino
Posts: 216
Joined: Fri Jun 11, 2010 8:40 am
Location: Italy
Contact:

Re: The TCEC...

Post by Carotino » Sat Nov 26, 2016 7:22 pm

Uri Blass wrote:
Carotino wrote:Adam, your mama did not teach you education? Why do you insult me?
Well, we fly over the peasants and talking about chess.

This is the variant used in the last game:

1.e4 e5 2Nc3 Nf6 3.Bc4 Nxe4 4.Qh5 Nd6 5.Bb3 Nc6? 6.Nb5! g6 7.Qf3 f5? 8.Qd5 Qf6 9.Nxc7+ Kd8 10.Nxa8 b6 11.Qf3 Bb7 12.h4 Bh6? 13.d3 f4?

Of course, Black (Houdini) lost the game.What sense does this thing?
Stockfish did not lose playing black with the same opening against Houdini.

I guess that the target of the openings is to help the stronger program to win.
With balanced opening there is a danger that we get 100 draws(and even if it is only 95 draws then it will be a disappointment for the spectators) so using unbalanced position may be the only way to see which engine is stronger because engines are almost unbeatable in balanced opening and I guess in some years we are going to have 100% draws in long time control games between top engines unless you choose opening like that opening when it is not clear if white win or if it is a draw.
Uri, in my opinion, the ideal solution would limit the opening book in two or three moves. The current engines (also seen the super-hardware used in TCEC) are able to deal with the first phase of the game autonomously.

Best solution for me:
a) Book composed of two or three moves;
c) limited time to 1 hour for each engine;
d) the presence of a human expert (IM or GM) who comments on moves and games.
Roberto

Lyudmil Tsvetkov
Posts: 6052
Joined: Tue Jun 12, 2012 10:41 am

Re: The TCEC...

Post by Lyudmil Tsvetkov » Sun Nov 27, 2016 7:41 am

beram wrote:
Modern Times wrote:Time control is way too long for me to even think about watching any games. Increasing the number of cores was a good opportunity to shorten the time control in my opinion. However you can look at the games afterwards if that is an issue I suppose.

What they have done with TCEC is incredibly impressive though, all credit to those involved.
I agree with you Ray and with Graham, that thx must go out to those involved with TCEC
On the other hand I share your criticism on the way too long time control

The opening choice i find interesting and varied enough, but the treshold for first move out of book could imo have been set slightly lower e.g. <0,6

To Ljudmil I wonna say be realistic, we have these two amazingly strong engines and we don't have any 300 ELO stronger other engine for now or within 6-8 years
what do you mean will not have within 6-8 years?

SF 8 is 300 elo stronger than SF 3, that took just 4 years.
whether one engine is going to overwhelmingly leave behind current tops or we will be witnessing the same dull almost-parity for another decade is debatable, of course, no one knows what the future hides.

TCEC openings are spicy, I grant that freely, but variety is not good, only 1.d4 and 1.e4 lines. Certainly this does not quite exhaust chess opening variety.

A reasonable suggesstion for next edition would be to use Clemens' Harry Schnaps book: excellent variety, very interesting starting positions, relatively short, non-biassed.

Post Reply