Carlsen plays the Drawmeister against Caruana and tricks him in the rapid games. Nakamura plays the Drawmeister against Caruana and tricks him in the rapid games. Do you recognize the pattern? Back in 1983 when the candidates match Smyslov vs. Hübner was drawn, the outcome was decided by a roulette ball. Now that is pretty lame one may argue, but it certainly has it’s advantages. Guess what, the Drawmeister becomes risky all of a sudden. If you have a 50% chance to get eliminated you can’t raise your overall chances by drawing games anymore. This may not be an ideal solution for a World Championship, because someone holds the title already, but it should work in regular tournaments.
After I read this article on chessbase.com I decided to take another look at this seemingly weird move. In order to discover the secret you have to click through every possible line, even if the engine considers it as dead lost. The key is once again to circumvent the horizon problem. Here is the surprising result, and since all the experts stay quiet on it, I assume that this is a world premiere:
It’s quite sad that Carlsen was such a favorite in the tie-break so that he could basically skip the entire classical part of the match.
This story has popped up on Reddit today. Four british players held AlphaZero to a draw with white by exploiting the fact that it only plays the “best” moves, which apparently leads to the Berlin. What a surprise, or not? Well, that’s exactly what this website is all about. White has at least a draw and he can force it in all major openings! With black it’s indeed a different story. They managed to hold in the King’s Indian where AlphaZero chose a line that was popularized by Sokolov in a match against Van Foreest in 2016.
The final position is a fortress, which is shows the second weakness of chess engines. Nice coincidence for a 2-game match.
Not only that engines only play the highest ranked move, which leads to horizon problems, they do not understand that a variation leads to a fortress early enough either, which is also a horizon problem. Humans on the other hand know what they are aiming for, so they can work their way backwards. Engines just make strong moves, while humans can use imagination to their advantage. For imagination to become a strength it requires some sort of knowledge though. If you don’t know what a fortress is, it doesn’t help to imagine the position where black is roughly a pawn down.
For the record: I can hold AlphaZero to a draw in Tic-Tac-Toe.
Here is some interesting data from the Leela Network. Page 23 from the supplementary data sheet of the AlphaZero Paper is also quite stunning. For example it tells us that this is a forced draw:
To put it mildly, I am not entirely convinced. There is a similarity to using an engine in correspondence chess. If you just play the top recommendation you won’t get far. It is well known that engines, even neuronal nets, have horizon issues that can be exploited.
Caruana’s team prepared a win, but Fabiano couldn’t convert it. That is what happened on the surface. In reality this game provides some insight in Caruana’s thought process. Instead of trying to create own threats, he was more concerned about stopping Carlsen’s threats. This works in positions with a static advantage, but apparently his advantage wasn’t static enough. Needless to say, he was a pawn down at that point. Note: Steinitz Rule 7.
If you check out the moves on Chessbomb, you will notice the absence of red moves. The key decision by Magnus was 44.Bxd5 of course. Desperate times call for desperate measures, but in reality the situation wasn’t desperate at all, it just required such play. The brilliant part about the decision to give the piece for three pawns was to realize that white could lose the queenside pawns and it’s still drawn because of the active king. He is the World Champion.
The best way for Caruana to deal with this emotional blow is to treat it as what it is, the past. You can only change the future.
It’s quite amazing that the Dragon is a super sharp opening, while with reversed colors it is quite harmless. Once again Caruana was prepared to defend a slightly worse but eventually drawn position. It looks to me that he doesn’t even try to fight for clear equality, but rather accepts a weakness or a slight disadvantage as long as he has analyzed this extensively. There is nothing else to say about it.
The Rossolimo leads to an asymmetrical structure where white has two reasonable pawn-breaks. In the first game we saw the f4-pawn-break, in the this game we see the b4-pawn-break. At first glance Caruana didn’t get anything, but analysis shows that things were not so easy. Apparently Carlsen’s 13…a5 was too optimistic.
Here is an interesting point: In the first game Magnus spoiled a huge advantage. In the second game he chickened out early and had to defend a worse position. In the third game he misplayed the position once again. So far Fabiano has made less big mistakes.
Play it safe!