TCEC Cup 9: future of chess?

[Jouni]

draw draw draw draw draw draw draw draw draw draw draw draw draw draw draw. No losses to top3 engines so far. Or Scorpio NN!

[Madeleine Birchfield]

The problem is the drawish opening book used in the cup. And the solution is to use a less drawish opening book, like what Joroen started doing in the TCEC superfinal starting season 18 and what Cato started doing in the TCEC Division Premier starting season 21.

[dkappe]

The problem is the drawish opening book used in the cup. And the solution is to use a less drawish opening book, like what Joroen started doing in the TCEC superfinal starting season 18 and what Cato started doing in the TCEC Division Premier starting season 21.

The problem? Then the solution is worse than the problem. The winner is the engine that can play unattractive positions that bear little resemblance to normal chess. TCEC DivP and SuFi are almost a variant that I’ll call Ubalanced Chess.

[Madeleine Birchfield]

The problem is the drawish opening book used in the cup. And the solution is to use a less drawish opening book, like what Joroen started doing in the TCEC superfinal starting season 18 and what Cato started doing in the TCEC Division Premier starting season 21.

The problem? Then the solution is worse than the problem. The winner is the engine that can play unattractive positions that bear little resemblance to normal chess. TCEC DivP and SuFi are almost a variant that I’ll call Ubalanced Chess.

Rating lists are also doing the same thing as well, using unbalanced positions to test for rating differences in engines. Stefan Pohl with his drawkiller opening book used in SPCC and Ed Schroeder with his gambit opening book used in GRL.

[ChickenLogic]

The problem is the drawish opening book used in the cup. And the solution is to use a less drawish opening book, like what Joroen started doing in the TCEC superfinal starting season 18 and what Cato started doing in the TCEC Division Premier starting season 21.

The problem? Then the solution is worse than the problem. The winner is the engine that can play unattractive positions that bear little resemblance to normal chess. TCEC DivP and SuFi are almost a variant that I’ll call Ubalanced Chess.

"Little resemblance of normal chess" LOL. Because humans would NEVER EVER go into unbalanced positions to outplay their opponent!

That's why we never praise Tal for going into unbalanced positions all the time. There is only chess regardless of "balance". We don't have to watch openings where we already know which line is going to be chosen by the engine just because you don't deem it to be 'real chess'.

[mvanthoor]

The problem? Then the solution is worse than the problem. The winner is the engine that can play unattractive positions that bear little resemblance to normal chess.

What is "normal chess" nowadays?

I follow Gotham Chess on YouTube, because of his series on tournaments in which he hopes to score enough rating points and norms to become a grandmaster. When I look at the chess openings that are played on those tournaments, between players in the 2100-2600 Elo range, I can only say that I'm flabbergasted. Some of the positions Levy calls "theory" and "main line" would have gotten you a fair scolding from your teacher in the 80's, because of "idiotic play." Most of those positions would probably have been deemed unplayable for one side or the other even 20 years ago, but nowadays, the computer has proven them to be fine.

I'm often seeing playing styles resulting from those positions that are completely alien to me. I've been playing chess for 35 years (off and on, sometimes with very long gaps where I didn't so much as touch a piece for 5-6 years), but the chess nowadays is completely different to the chess I grew up with.

"Normal" now isn't the same as "normal" back then.

The computer also has murdered everything we know about Go. Many Joseki (= "standard move sequences") that where deemed to be "out-analyzed" where proven to suddenly be losing for one of the players, and some theories about "how to play in situation X or Y" went out the window completely. AlphaGo has also often shown that "You can't do X in this situation" is untrue; AlphaGo did it X, and proven it to be at least equal, and sometimes winning.

[Modern Times]

The problem is the drawish opening book used in the cup. And the solution is to use a less drawish opening book, like what Joroen started doing in the TCEC superfinal starting season 18 and what Cato started doing in the TCEC Division Premier starting season 21.

The problem? Then the solution is worse than the problem. The winner is the engine that can play unattractive positions that bear little resemblance to normal chess. TCEC DivP and SuFi are almost a variant that I’ll call Ubalanced Chess.

Totally agree.

[dkappe]

I follow Gotham Chess on YouTube, because of his series on tournaments in which he hopes to score enough rating points and norms to become a grandmaster. When I look at the chess openings that are played on those tournaments, between players in the 2100-2600 Elo range, I can only say that I'm flabbergasted. Some of the positions Levy calls "theory" and "main line" would have gotten you a fair scolding from your teacher in the 80's, because of "idiotic play." Most of those positions would probably have been deemed unplayable for one side or the other even 20 years ago, but nowadays, the computer has proven them to be fine.

I also follow Gotham Chess. Highly recommended. I’d still argue that though engines have strongly influenced openings, that’s still far away from the outright crap that gets played in these imbalanced openings. I enjoy a good pawn odds game too, but is that chess?

[Madeleine Birchfield]

I enjoy a good pawn odds game too, but is that chess?

The latest VSOB somebody submitted two openings where white gave black h-pawn odds and a-pawn odds respectively and the engines playing (think it was Leela vs Stoofvlees for the h-pawn odds and Scorpio vs Ethereal for the a-pawn odds) drew the four games if I remember correctly.

[lkaufman]

I enjoy a good pawn odds game too, but is that chess?

The latest VSOB somebody submitted two openings where white gave black h-pawn odds and a-pawn odds respectively and the engines playing (think it was Leela vs Stoofvlees for the h-pawn odds and Scorpio vs Ethereal for the a-pawn odds) drew the four games if I remember correctly.

That's not surprising, because with either "a2" or "h2" pawn odds, White has both the first move and an immediately active rook for the pawn, which is nearly enough comp; White is probably not much worse than Black is in normal chess. Edge pawn odds are basically half pawn odds.