Not logged inRybka Chess Community Forum
Up Topic The Rybka Lounge / Computer Chess / TCEC 19 superfinal openings available
- - By Venator (Silver) Date 2020-10-17 07:34 Upvotes 2
Here is my selection of 50 openings for the TCEC 19 superfinal, won by Stockfish by 54.5-45.5 vs Leela Chess Zero:
Attachment: TCEC19_NoomenSelect.pgn (12k)
Parent - By Venator (Silver) Date 2020-10-17 08:53
All 100 games of the superfinal, without comments/main lines/evals:
Parent - - By Venator (Silver) Date 2020-10-17 10:45
Some interesting statistics:

In the Leela wins, the book exit evaluations by Leela were: 0.43 / 0.54 / 0.50 / 0.46 / 0.66 / 0.48 / 0.56 / 0.52 / 0.63

I.e. the lowest book exit that led to a win was 0.43 (which is quite high by Leela standards). All other white Leela games that had lower exits ended in a draw.

In the SF wins, the book exit evaluations by SF were: 1.14 / 0.99 / 0.77 / 0.63 / 0.98 / 1.39 / 1.06 / 1.19 / 1.14 / 1.20 / 0.70 / 0.66 / 1.38 / 1.05 / 0.91 / 1.27 / 1.00 / 0.85  (note: with contempt=24)

I.e. the lowest book exit that led to a win was 0.63. All other white SF games that had lower exits ended in a draw.

Conclusions:

1. All balanced lines, slightly unbalanced lines and slightly dubious lines ended in draws => the draw rate for such lines is 100%;
2. The ONLY wins in this superfinal were scored in (very) dubious lines.

Of course it's a small sample size, nevertheless I think these numbers are quite telling.
Parent - - By MrKris (***) Date 2020-10-17 21:26
Nice to see the actual data confirming what many, I think, suspected.
Thank you for compiling it.



P.S. Sorry, I do not have time to search for the link to this, which is maybe somewhat related.
To paraphrase from my memory:

     The ICCF World Championship [a large round-robin] is approaching 100% draws.
     This was not the case decades ago.

Parent - By Venator (Silver) Date 2020-10-18 04:42 Edited 2020-10-18 04:57
Nice to see the actual data confirming what many, I think, suspected.

Indeed. Computer chess is approaching a 100% draw rate when using balanced or slightly unbalanced openings in the TCEC settings (i.e. very fast hardware and a long time control). Even slightly more dubious lines led to 100% draws.

Still, many people are calling for balanced lines. I think they have no idea what they are asking for. The draw rate would be like 96-98%, with 90% of the games being uninteresting and dancing around 0.00 scores. It's the easiest way to chase away the spectators and to get a completely random result, where the winner is lucky to win one game extra because of one big mistake.

It's getting progressively more difficult to find lines that lead to a 1.5-0.5 result, especially now both finalists use a NN evaluation. In S9 through S15 I could pick openings with book exits under 0.50 (in SF terms) to get exciting games, a reasonable amount of wins and sufficient 1.5-0.5 pairs. Unfortunately that won't work anymore, which I found out in S16. The number of 1.5-0.5 pairs has also been steadily declining, even with more dubious openings.

Anyway, there were lots of interesting theoretical discussions in the TCEC 19 superfinal, which will make their way to theoretical articles.
Parent - - By anst (*) Date 2020-10-18 08:21
Thank you very much for your dedicated work for so many years !
Sad to say these matches show that in many ways superchess is now more or less over. In addition to what you write I have noticed that already when exiting from the opening the super engines see that it is a draw and the rest of the game is really unnecessary.
Your analysis should be put into a more formal paper I think with a catchy title like "The end og chess" :(

Perhaps we have to pursue other arenas to get unexpected results - Giro d'Italia is on !

Best regards
anst
Parent - By Venator (Silver) Date 2020-10-18 09:03
Thank you.

In addition to what you write I have noticed that already when exiting from the opening the super engines see that it is a draw and the rest of the game is really unnecessary.

You are absolutely correct. And you will get 90 of such games (or even more?) under TCEC conditions, when using balanced openings. Even +0.20 or +0.30 book exits will quickly converge to 0.00 evaluations. And as soon as the evaluation is 0.00, the rest of the game is not interesting anymore.
Up Topic The Rybka Lounge / Computer Chess / TCEC 19 superfinal openings available

Powered by mwForum 2.27.4 © 1999-2012 Markus Wichitill