Alpha Zero Proves That King and Two Knights vs. King Can Be Won in 73 Moves
Is there anything that engines can't do? Sigh.
Is there anything that engines can't do? Sigh.
Every week ChessLecture.com makes a free video available (which is available for two weeks); this week it's another one of my videos - an oldie (recorded in 2011) discussing the game Nowak-Pachman, Solingen 1968. It's an instructive victory by Black in a same-colored bishop ending, showing the value of a space advantage in what might otherwise seem a very drawish endgame.
The video is here, and available for free. (If you don't have a membership, you can create a free account.)
Wesley So isn't as active in the show as one might like, but it's still nice to see him participate in Karsten Mueller's presentation of some of So's interesting, mostly recent endgames, here.
It was an exciting day of tiebreaks, though it was disappointing that only one match made it past the two 25-minute games, and it was settled in the 10' + 10" round. We need to see at least one Armageddon game before the tournament ends!
Anyway, to the round. Peter Svidler had the easiest time of it, beating Bu Xiangzhi 2-0. In the first game, Svidler won with Black after Bu got tangled up in the center. White tried to bail out with an exchange sacrifice, and it almost worked. Bu was about to esacape until he played 40.Ra5??, walking into a lethal self-pin. Walking into mate in one on the next move didn't help, but the damage had already been done - even 41...Rb4 would have done the job. In the second game, Bu tried the Dutch, hoping for a complicated position, but when he met the Improved Lisitsyn Gambit by turning the game into a Philidor Counter-Gambit he got in trouble - fast. He was already clearly worse by move 7 (maybe by move 5, but let's be generous), and after a huge error on move 9 he was completely lost. Svidler may not have played in the most incisive way, but he didn't have to, and he coasted to victory.
Wesley So was also a smooth winner, outplaying Baadur Jobava in their first game with the white pieces, demonstrating the power of the bishop pair (and later of bishop vs. knight) to grind out a victory. Game two was an "I'll make you an offer you can't refuse" draw: So was better from early on, and could have played for a win had he needed to. Instead, he allowed Jobava to draw by repetition in a position where he was still better, but the problem for Jobava was that varying from the repetition would lose on the spot.
In one of the matches featuring underdogs, Evgeniy Najer held an edge in his white game with Richard Rapport until his ill-advised 23.Bxh6, which should have been met by 23...Rxf3. For a while after that Najer had good chances to win, but Rapport gradually clawed his way back to equality and a draw. The second game was completely crazy, and Rapport handled the complications much better than Najer to win deservedly. There was one big hiccup near the end, however. 45.Rb7+ followed by 46.Re2 won comfortably, but after his 45.Rb6? Najer had 45...Re1+ first, and only after 46.Kg2 was 46...Nb4 correct. In this case he would have equalized. Now White can't play Re2, and if he takes on a5 Black has an immediate perpetual with his rook going to e2, e1 and/or e3, as needed.
In the other battle of the underdogs, Vladimir Fedoseev defeated Maxim Rodshtein 2-0, though unlike Svidler's 2-0 victory it wasn't easy. First of all, it's a mystery why Rodshtein didn't play 37...Qxc3 in the first game, leading to a dead draw after 38.Rxc3 Bxf2 39.Nxe6 Rxc3 40.Bxc3 fxe6. Even after 37...Bxc7 39.Bxc7, trading queens would have given him excellent drawing chances in the opposite-colored bishop ending. The draw wouldn't be guaranteed on account of the rooks, but keeping queens on as well made it harder, not easier, for him to defend. Eventually the queens came off, but under more favorable circumstances for Fedoseev. It still wasn't easy for White to win until Rodshtein's 69...Kf7, allowing White to play 70.Rf8+ and 71.Rf6. After that, the conversion was routine. Rodshtein did a great job of creating a complicated mess in game two, and he had good chances to win as soon as the early middlegame. The game went back and forth, and Rodshtein missed a very good chance on move 33, when taking on b5 followed by d6 would give him a winning advantage. From there on, he played too passively, and Fedoseev took over the initiative. White had to play 41.Bg2 to stay alive, and after missing that chance he resigned three moves later.
Finally, in a match that would have been better as a semi-final or even a final, Maxime Vachier-Lagrave and Alexander Grischuk had a heavyweight battle in keeping with their ratings. They drew the 25-minute games, and saved the best for last. In the first 10-minute game, Grischuk's attempt to solve his strategic problems with tactics failed. In particular, 28...Rg6, going for counterplay, was strongly met by a great pawn sacrifice from MVL. From 30 to the end of the game, Vachier-Lagrave blew his opponent off the board with one threat after another in a great display of the power of the initiative. The second game was a battle between the initiative - again, on MVL's side - and static goods. Grischuk's 10.Bxc6 wrecked Black's queenside structure, but at the cost of the bishop pair, weak light squares, and a few moves later, a badly sidelined queen. Vachier-Lagrave found a great exchange sac, but misplayed it a few moves later and wound up in an inferior ending. After two further inaccuracies, he wound up in a lost ending with bishop and pawn against Grischuk's rook and pawn. Grischuk made a very serious practical error when he didn't play 44.h3, after which proving a win with hardly any time on his clock was as good as impossible, and MVL advanced to the fifth round.
The games are here, but I've only annotated the second Svidler-Bu Xiangzhi game, along with the two MVL-Grischuk 10-minute games.
Tomorrow the quarter-finals begin, with these pairings (in bracket order): Svidler - Vachier-Lagrave, Ivanchuk - Aronian, So - Fedoseev, Rapport - Ding Liren.
Will Svidler continue his question to reach his fourth consecutive Candidates event? (Admittedly, once he was the organizer's wildcard pick, but the other two times he qualifed through the World Cup.) Or will Vachier-Lagrave stay alive as he hopes to reach the Candidates for the first time in his career? Can Ivanchuk survive the top remaining seed, Aronian, and show that his glory days are still going? And will the young upstarts Fedoseev and Rapport (22 and 21 years old, respectively) be put in their place by their elderly opponents (So and Ding Liren; 23 and 24 years old, respectively)?
There were three decisive games today, and there are three players advancing to round 5, but there isn't a one-to-one correlation between the two "threes". Ding Liren defeated Wang Hao in a good game with White in a Catalan, but if Wang Hao had known about an earlier game - or simply found the right idea on move 22 - the game probably would have finished in a draw, and they'd be off to tomorrow's tiebreaks.
Levon Aronian also won, defeating Daniil Dubov in a long game. Aronian reached a theoretically won ending, and while he had time at the start to figure out how to win it, he didn't hit on the right plan. Over the course of the next many moves, he even allowed Dubov numerous chances to draw, but Dubov - who had the time and ability to work out his drawing opportunities - thought it was the better strategy to keep blitzing Aronian. It backfired. It took Aronian seemingly forever, but around 40 moves later than he could have won, he finally hit on the right strategy - though he still managed to give Dubov one more (missed) drawing chance after that. Should Dubov have taken his time? The problem is that if he did, at a moment when he didn't have a draw, it could very well have given Aronian the chance to work out the winning plan. So I think Dubov was generally right to blitz - given his correct assumption that the ending was generally lost. But there were several positions where it looked like he could have an escape, and that's where it would have made sense to slow down and look. It's a risk, but there I think it's worth taking. Anyway, he's out, and Aronian advances.
The day's third winner was Maxim Rodshtein, who leveled his match with Vladimir Fedoseev. The game was an odd echo of the previous day's game: both won with Black after creating complications starting with a dubious ...g5 pawn sac. Fedoseev seemed too intent on playing for a draw - certainly in the opening - and it allowed Rodshtein to make lots of trouble for him. His reward: tiebreaks tomorrow.
The third player to advance is Vassily Ivanchuk, who was beating Anish Giri today, too, but he made Giri an offer he couldn't refuse: allow an immediate repetition or be dead lost. Giri chose to keep most of his rating points, and called it a tournament. Ivanchuk, meanwhile, will play Aronian in the quarter-finals in the only match that's set so far.
The other four games finished in draws and will result in tiebreaks. Alexander Grischuk vs. Maxime Vachier-Lagrave was a 13-move draw; apparently Grischuk is reverting to his gruesome but effective strategy from Kazan Candidates matches a few years ago, where he would draw all his classical games with White without a fight and then hang on desperately with Black, aiming to reach the rapid and blitz tiebreaks.
Bu Xiangzhi vs. Peter Svidler was also a short draw, but this doesn't seem to have been by design. Bu was outplayed in the opening, and was pulling on the emergency brake before things got out of hand.
Baadur Jobava outplayed Wesley So and had him on the ropes, but So saved the game by creating a fortress in the ending.
Finally, Evgeny Najer and Richard Rapport had a hard-fought draw. It looks like Najer generally had the better chances, but Rapport was never at death's door.
Games, with mostly brief comments, here.
There's a nice video series by Mark Dvoretsky (hosted by Jan Gustafsson) on endgame play (on Chess24) that's worth your while, especially (as is generally the case with Dvoretsky's material) if you're at least 1800-2000 in strength. (Of course you can learn plenty from him even if you're not yet of that strength, but he does pitch his material higher rather than lower.) His elocution could be better, but the material is excellent.
Wednesday was a rest day for the participants in the Sinquefield Cup, and before that was round 5. In the two previous rounds all the games were drawn, and the first four (of five) games to finish on Tuesday also finished peacefully. One game remained, between Veselin Topalov and Ding Liren, and although Topalov was winning earlier and still had some advantage, it seemed to be headed for a draw as well. But that's when it got interesting, as you can see for yourself.
Round 6 is today, with these pairings:
Test your endgame knowledge and skill here, and make a contribution to science. (HT: Hylen) If you're game, leave a comment with your score and your rating; I'll kick things off with the first comment.
G. C. Van Perlo, Van Perlo's Endgame Tactics: A Comprehensive Guide to the Sunny Side of Chess Endgames(New, Improved and Expanded Edition) (New in Chess, 2014). 607 pp., $34.95/$29.95. Reviewed by Dennis Monokroussos.
This fascinating and delightful book may not be the "best" endgame work ever written, but it's probably the most enjoyable book on the subject. The late G.C. van Perlo, who passed away in 2010, was a correspondence chess grandmaster and a terrific writer, and the first edition of Endgame Tactics won lots of awards when it came out in 2006. There were minor changes in the next two editions - corrections to mistaken analyses - but this latest edition has some significant additions - it's 25% bigger, according to the back cover. Van Perlo had finished a new book on rook endings that was similar in style to what he had done in the rook endings section of Endgame Tactics. The book hadn't yet been published, so it was decided to add it here as a new and large chapter to his already sizable masterwork.
For those who are new to the book, it consists of a huge number of endgame positions which mostly involve some tactical nicety or niceties (thus the book's title) - fine points which are often missed. A huge part of the book's charm comes from the author's writing. Van Perlo writes with a good deal of punchy humor and puts a psychological twist on many of the fragments, and both work well given the huge number of heartbreaking failures displayed in the book.
In all there are a whopping 1368 fragments, 75 of which are official exercises included in the material new to this edition. As in traditional endgame texts, the positions are divided by material (pawn endings, queen endings, rook endings, minor piece endings and so on, with further subdivisions as appropriate) and then often further divided by themes.
Anyhow, the book has been reviewed and praised many times over the past eight years, so rather than gild the lily and bring coal to Newcastle I'll stop here and join the chorus. It's a great book, and one practically every chess player of whatever strength can and will enjoy, whether he is a fan of endgames or not. The book can be used to help learn endgames, to practice tactics, and even just for the sheer pleasure of seeing beautiful tactical ideas and/or for the occasion guilty pleasure of schadenfreude.
Recommended to all chess players.
In my summary post on Cap d'Agde, I mentioned a long and fascinating ending between Anatoly Karpov and Mariya Muzychuk from their semi-final match. Here it is, with my annotations.