seems they were right. AlphaZero winning 28 and drawing 72 against Stockfish is pretty damn impressive. That would translate to roughly 150 Elo points I believe.

I for one welcome our new robot chess overlords

slow pony

glad this was finally done by the alphago team as it seemed like a half day project

ICYMI: Liv Boeree has recently started playing blitz almost daily on chess.com.

I just thought ya had a sweet tooth.

One of the more interesting things about AlphaZero's conquest of chess was seeing what openings it liked and didn't like as it got better and better. Stolen from chess24:



Originally it liked the Caro-Kann, but threw it in the trash as it got better. It now likes the English, the Queen's Gambit and other quiet d4 openings, and the Berlin as Black. So like every other super-GM, then.

It's a shame only 10 games have been made public,

I really want to see one of the 3 games where AlphaZero won as black

Found one... game starts at 2:30

Some moves are just way beyond my comprehension, I wouldn't even consider some of the moves that happen in them

Black's position looked bad to me in that game pretty much right up until the point where it started strangling White to death.

Sick game.

Watched game 2 just now. Another Black win. Positional murder. Very nice game.

It turns out there were 1300 games, not just the 100. The other 1200 were playing specific openings.

Hopefully they are all released!

I'm... 98% convinced this thing is our new overlord, but that time control was sketchy.

Naka fuming about no opening book is funny too. Although I think Svidler even asked if that's really true, how are the 10 example games playing actual opening theory without Stockfish having some sort of opening book. Sure Alphazero may have learned common openings but Stockfish needs the book.

Do you mean the white side of the Caro-Kann, as in originally it liked playing white against the Caro-Kann then not so much soon after?

I'm late to the game. Chess lover. Please add me to chess.com. My username is atcherryfindom

Thanks.

I'm not a troll, I'm a cat

and thank you, that is sort of what I looked like as a baby

An example of optimism at its finest tonight. It's White to move and my scholastic opponent appears to be completely busted:



Ten moves later, I appear to have him badly tied up in addition to owning the bishop pair and an extra pawn. Engines agree that White is in serious trouble:



But here is the problem... I am way down on time after spending too many precious minutes searching for a kill shot that wasn't there for the past 10 moves. After a couple of inaccuracies and blundering a pawn, I am suddenly the one praying for a draw. I am in massive time trouble while he has all the time in the world on his clock. Thankfully, he offers a draw and I insta-accept. I have seen him do this quite a few times with other opponents. He knows that he has an excellent chance to grind me to death and win, but that is not his nature at this point in his life. He's just too nice to to do that to me.

In the post-mortem, he plays through the game with me explaining why his position was totally fine all along in his mind. He is, of course, objectively wrong. But I have zero doubt that he was relaxed and confident all along due to his extremely optimistic view of a busted position. "Black's pawns are super weak, so he isn't really a pawn up in my mind!" "I am well on my way to setting up such a strong blockade that Black cannot improve his position!" "Black's light-squared bishop is horrible while my knight dominates the center!" And so on.

After we finish the post-mortem, he flashes a mischievous smile and adds "But to be honest, I bet the engine hates my position!" I laugh and add that the engine will likely laugh at my moves as well.

Defending a busted position is one of the most difficult and unpleasant tasks in chess. Many 1800ish players throw a hail mary or two and head home when they end up in such a spot. But my opponent found a way to make the most of what he had. Sure, he is a talented young player who is well on his way to expert. But he defended himself like a true master tonight, leaving me in awe of his heart and character.

Psychology is very important. Although in this specific case I'd say it's a bit more that he cannot properly evaluate a position and/or extreme optimism vs. he is good at hunkering down and grinding a bad position. If he was good at grinding he'd have kept playing.

Extreme optimism is good in many games when playing chess but it also is quite harmful since you can't really ever objective evaluate a position, which is kind of important. On the other hand, Kramnik seems to do fine with it.

Speaking of over optimism, in one analysis session, my friend's student suggested a move that led to a big attack for his opponent. When shown what was about to happen, the kid said "I can survive this". My friend replied "Yeah, and you could survive a bear mauling, but why would you want to?"

It's probably a "had to have been there" sort of thing, but he definitely offered the draw out of kindness. I know that I was visibly sad and exhausted. And I have seen him do it too many times to think otherwise. If you remember the scene from "Searching for Bobby Fischer" where Josh sees that he has a won pawn ending, but offers his more sadistic opponent a draw out of the kindness of his heart, it is that sort of thing. Jeremy Silman argues that the key to winning long, grindy endgames is to take sadistic pleasure in slowly torturing your opponent to death. He just doesn't like to do that yet. I am sure he will change his mind in big tournaments and when he starts approaching master.

He definitely has the skills to grind and win brutal endgames. I have played him 4 times. The first time, he turned a draw into a win. The second time, he turned a loss into a draw. The third time, I had to play way above my head to convert my opening tactic into a win (was down to increment for the last 20 moves or so because he defended his terrible position so well). And this time, he converted an objectively "lost" position into a draw and could have easily played for a win. He is an excellent defender.

I should also add that during our post-mortem, multiple veterans of the chess club dropped by the table to make fun of him for his extremely optimistic view of his chances. He was unaffected. He is extremely well-coached. His father and little brother are always smiling and joking around about his antics like he does. My takeaway from the Magnus documentary was that Magnus got so good because he loves chess so much that he always wanted to study and play growing up. I think Magnus said that the day he stops having fun will be the day he retires.

So I was on a 7-0-1 undefeated streak until I ran into this buzzsaw of an attack last night. My opponent is a senior player rated around 1870 who brutally reminded me that even "old guys" can play like maniacs and get away with it! A truly beautiful attack to the point that even the engines get confused. The engine may say that I am well ahead at certain points, but it usually peters out into a draw with best play if I continue to follow the engine's suggestions:

https://lichess.org/study/psc7OYOi

Note to self: take this guy's attacks extremely seriously and consider declining similar exchange sacrifices in the future. That h2 square was extremely difficult to defend once I lost my f3 knight!

Nice king walk.

Yeah. That is tough. Are you familiar with that opening vs. the Chigorin. It could be that is a very standard attacking opportunity in that spot. It seems to me just avoiding it entirely is a good idea. E.g. castling later or something. I assume there is some theory on it but do not play either side of this so have no real idea. I'd be tempted to just play h4 vs. h5 even though it makes the g4 square available. Since after h4 it just seems Black will get a powerful and relatively direct attack no matter what you do. And it's not that hard to play the Black side of this tbh, he didn't have to make any really tricky moves or anything. That's why I might prefer just avoiding the entire thing altogether.