Speaking of engines, has anyone kept up with what happened after they turned Alpha Go on to chess. Is that the strongest engine now?
Didnât read the full article and donât even play anymore so what do I know but people have complained forever that creativity was lacking, games were boring, and there were too many draws. Psychology and gamesmanship have always been part of play. Strong chess computers have been around for decades. Cheating isnât new; Niemann sure didnât invent it. You can compare chess to poker if you want but these factors donât add up to a good case that chess is a lot more like poker now than it used to be in the good old days.
AlphaZero may or may not have been stronger than Stockfish back when it was first trained, but Stockfish also now has a neural net eval function since Stockfish 12 (a shallow neural net unlike the deep neural net of AlphaZero) and Stockfish is definitely stronger now. Leela Chess Zero is an open-source copy of AlphaZero, basically, and is both stronger than AlphaZero was and weaker than current Stockfish. Itâs not surprising that AlphaZero, which is a general-purpose neural network capable of learning many different games, is not as strong as a neural net specifically designed to be a good fit for chess. But AlphaZero âwonâ in the sense that hand-crafted eval functions are a thing of the past; every serious engine uses neural net eval now.
Itâs hard for any engine to be better than Stockfish for long, since Stockfish simply copies successful ideas and has an army of programmers ready to quickly implement proven techniques.
A more interesting âIs Chess The New Poker?â article probably would have been about the explosion in chess activity after the pandemic collided with Netflixâs Queenâs Gambit series. The increase in the last couple of years in the popularity of chess has some parallels to the early 2000s in poker. The story of chess.comâs monopolizing behavior, increased on line chess traffic, and the increase in popularity of popular chess streamers would actually be kind of interesting.
2 Likes
is there a sense if there is a rating ceiling for computers? like yeah stockfish X+1 could be marginally better than stockfish X, but it translates to maybe half of one win out of 100, and the rest are draws, regardless of opening line.
Chess and poker are more alike than they are different for sure. The demise of both due to AI has been predicted for a generation. What the article fails to mention though is that despite the fears, theyâre both arguably more popular than 10 years ago! The Eval bar has done for chess what hole card cams and EV %s did for poker two decades ago. It turns out that knowing the GTO solution in a situation makes viewing more fun, not less. Seeing top players make objectively bad decisions just makes them more relatable to us. I mean yea, sucks for the true endbosses that their edges are miniscule and they have to go deep into the lab to find them. But in a world of chess960, 5 card plo, and shot clocks, the end of competition looks farther away than ever.
2 Likes
Well there has to be at some point. I think thereâs still a fair bit of theoretical room for improvement. Rybka seemed like a god in its day but itâs hopeless compared to modern engines.
Gonna need to update that Atlantic article. This incident makes it even clearer that chess is poker.
Magnus is just embarrassing himself now, if you donât want to play the guy just forfeit ahead of time.
Some kind of sanction is warranted. No world championship defense. Ditching tournaments. Rage quitting games. If he doesnât want to play, thatâs fine. But this is bullshit.
I think a pregame forfeit gets you dqâed from the whole tourney. But seriously,this is dumb
Yeah he needs to either make accusations or stop being a diva.
looking over the results table, magnus is a favorite in his remaining games, and he has only 2 points to make up to and possibly still win the tournament. niemann had a good day one, but besides the free points, he was 0.5/3 in his other games on day two.
if chess is like pokker, then this is obviously trying to embarrass niemann, and perhaps it worked.
i donât know the order of the games as played, but according to chess.com it looks like niemann lost to a 2450 rated Erigaisi right before playing magnus. maybe, just maybe, magnus saw that and decided he could still win while throwing this one game
That must be rapid rating . Erigaisi is over 2700 in classical and considered to have a good chance to break 2800
itâs absolutely rapid rating. magnus with 2843 in rapid. so sick 
Is this like a chess version of Mike Postle where only people who understand chess at higher than a basic level can âseeâ the obvious cheating which is, I guess you could say, circumstantial because there isnât a smoking gun confession or whatever? The part where Chess.com said âyeah we have more data that says you cheated a lot more than youâve admitted, feel free to correct the recordâ seems pretty damning, even if heâs not cheating OTB?
This video is the most compelling case Iâve seen
https://twitter.com/MikeMcDonald89/status/1569392703133163523
Yeah I was just watching that one but donât know enough chess to understand the context. If Iâm hearing it right, a ~20 CP mean loss would be plausible for a 2700-caliber player, but then this shift to sub-8 CP is not a reasonable thing to expect from humans over, say, an entire tournament? And also this shift coincides with tournaments that are more influential on ranking?
This guy is the most prominent expert on cheating in chess. He was interviewed on the broadcast today. He said heâs examined all of hansâs over the board games since maybe 2019 and found no evidence of cheating. In a tournament Hans won in Havana, this guy found that Hans performed slightly under expectation, but still won because the other players played even more under expectation. Yet people point to that tournament as evidence that heâs a cheater