Computer Grob Fail

by admin on April 21, 2018

So far this year, I haven’t written a single post about a game against the computer. I’ve been avoiding that on purpose, because I overdid it last year. But two of the things that give me the most satisfaction in life are beating the computer and beating the Grob. When the two come together, and it’s an entertaining game with three pawn sacrifices and a king hunt, how can I not write about it?

This was played yesterday against Shredder, with its rating set to 2220 (far below its top level, which is 2600, but strong enough to usually beat me).

Shredder – Dana
1. g4 d5 2. h3 e5 3. Bg2 Ne7

This is quite unusual, with only six games in Chessbase, compared to 515 games with the “normal” move 3. … c6. I’m not sure why there should be such a big difference; in many cases they can transpose.

4. d4 e4 5. c4 c6 6. Nc3 Ng6

Very provocative. It’s trying to put pressure on my center, and I remove a defender of the center. Of course, the reason is that I want to exploit the weaknesses on its kingside. To which Shredder replies: Weaknesses? What weaknesses?

7. cd cd 8. f3?! …

Position after 8. f3. Black to move.

FEN: rnbqkb1r/pp3ppp/6n1/3p4/3Pp1P1/2N2P1P/PP2P1B1/R1BQK1NR b KQkq – 0 8

If you asked me for one move that crystallizes the difference between computer chess and human chess, this would be it. No human would ever, ever consider such a move, which weakens an already porous kingside. But Shredder cares only about blowing up my center; and it correctly realizes that White has nothing to fear from 8. … Qh4+ 9. Kf1 ef, when either 10. Bxf3 or 10. Nxf3 Qd8 11. Qb3 win a pawn, with no real compensation for Black.

I’ve had so many games against the computer where it plays moves that, according to human intuition, should not be playable, and yet it wriggles out of trouble and even comes out ahead. It’s really nice, just for a change, to have a game where human intuition beats the brute-force calculating ability of the computer.

8. … Nh4! 9. Bf1 Bd6 10. Be3?! …

Another computer move. I think almost any human would try to close the diagonal with 10. f4. But Shredder doesn’t care. Rybka thinks that White should play 10. fe Bg3+ 11. Kd2 de with only a very slight (0.2-pawn) edge for Black, but I would have been happy to play this position.

10. … Bg3+ 11. Bf2 Qd6 12. e3 …

Practically the first “normal” move Shredder has played. The variation 12. fe Bxf2+ 13. Kxf2 de 14. e3 O-O would transpose to the game.

12. … O-O

I took quite a bit of time on this move. I think that 12. … f5 looks safer and probably I would play that move in a tournament game, but this was just a fun 10-minute game, and I thought, “What the hell. I just want to get my pieces out.” In addition, I wanted to see whether Shredder would be tempted to take the pawn sac.

Position after 12. … O-O. White to move.

FEN: rnb2rk1/pp3ppp/3q4/3p4/3Pp1Pn/2N1PPbP/PP3B2/R2QKBNR w KQ – 0 13

13. fe?! …

Hunch confirmed. This move is very consistent with Shredder’s previous moves. Safety is a foreign concept to Shredder; it just sees that it can win a nice center pawn.

Again, I think that a human player would avoid this move on principle and try to close lines and force Black to trade bishops with 13. f4. Even better (according to Rybka) is 13. Qe2 with the idea of playing 14. f4 next move; the point of 13. Qe2 is to allow White to reply to … Bxf2 with … Qxf2. Rybka evaluates the position as dead equal.

This goes to show that not all computer programs are the same. In this game I was definitely able to exploit Shredder’s hyper-aggressive tendencies.

13. … Bxf2+ 14. Kxf2 de 15. Nxe4 …

In for a penny, in for a pound.

15. … Qd5?!

My biggest misstep of this game. The queen turns out to be somewhat vulnerable here, because after … f5 it is on the same exposed diagonal as the king. The move 15. … Qe7! is just as threatening and doesn’t have the same liabilities.

16. Qc2 Be6 17. Kg3 …

Brave but also quite logical; White needs to get the king off the f-file.

17. … f5! 18. Kxh4? …

But now Shredder takes bravery one step too far. White has to try 18. Ng5, and Black’s best reply according to Rybka is 18. … Rc8 19. Qf2 (or Qh2), with an approximately equal position.

18. … fe 19. b3 Qd6!

Sacrificing a second pawn — but tightening the mating net around White’s king.

20. Qxe4 …

The critical moment of the game. How should Black proceed?

Position after 20. Qxe4. Black to move.

FEN: rn3rk1/pp4pp/3qb3/8/3PQ1PK/1P2P2P/P7/R4BNR b – – 0 20

I won’t lie to you. For a minute after White played 20. Qxe4 I thought that I had screwed up. The reason is that my intention here was to play 20. … Bd5, and at the last minute I realized that 21. Qxd5+! Qxd5 22. Bc4 squashes Black’s attack flatter than a pancake. Also, on most other “reasonable” Black moves, such as 20. … Nc6 or 20. … h6, 21. Qxe6+ basically ends the game.

So Black has to strike fast, and I started looking at 20. … g5+ almost out of desperation. And then it dawned on me — holy crap, it works! It’s always a bit of a surprise when the computer allows you to play a tactical shot like this, and I’m sure that it has to do partly with the fact that I’m playing Shredder at a rating of 2220, not its maximum rating.

This is actually a good argument for playing the computer at less than its maximum strength, because at maximum strength you get used to the fact that it never overlooks a tactical combination. At some point, you get out of the habit of even considering those possibilities, which is bad because your human opponents definitely will miss tactical shots and you have to practice spotting them and pouncing on them.

20. … g5+! 21. Kh5 …

It declines the sacrifice because 21. Kxg5 would allow 21. … Qd8+ 22. Kh5 Bf7+ 23. Kh6 Qh4 mate. This is quite ironic, because when White plays 1. g4 and 8. f3, you expect him to get mated by a queen check on h4 — but not quite in this fashion!

21. … Bf7+ 22. Kxg5 Qd8+ 23. Kf4 Bg6+ White resigns.

White has to give up the queen, and its king will still be exposed to a lethal attack. A rather amusing final position, because 4 of White’s 5 pieces are still on their original squares, and 3 of Black’s 5 pieces are also still on their original squares.

I hope you appreciated this trip into the bizarre world of computer chess!


Print Friendly


Eagles Skewer Pandas in Championship Match

by admin on April 8, 2018

Season two of the PRO Chess League has now ended, and the Armenia Eagles are the champions! In fact, the championship match went exactly as Hikaru Nakamura predicted. He said in an interview before the match that he expected an 8-8 tie and then expected Armenia to win in tiebreaks because of their superior blitz chess. (The tiebreak consists of 3-minute games with a 1-second time increment.) And that’s exactly what happened.

Going into the last round of the match, the Chengdu Pandas has a 6½-5½ lead, and in the early stages of the last round things looked very promising for them. Armenia’s fourth board, Artak Manukyan, wildly sacrificed two pawns for dubious counterplay and looked as if he was on the way to defeat. That would have forced Armenia to make up a two-point deficit, and none of the other three games looked all that promising for them to win.

Black to move.

FEN: r1b1r1k1/pp3ppp/6n1/3q4/1B1pnP2/3B3P/2P3PK/R2QNR2 b – – 0 1

But in the position above, instead of  playing the safe 1. … b6, planning to move the bishop to b7 with a nice battery on the long diagonal, China’s fourth board Chu Ruotong played the suicidal 1. … Bf5?? Manukyan pounced on his opportunity with 2. Ra5, skewering the queen and bishop … and then he offered a draw!

In fact, he would have been winning after 2. … Qd7 3. g4, trapping Black’s bishop.  Why did he offer a draw? It’s all psychology. When you’ve been in a losing position for a while, just hoping to escape somehow with a draw, and then your opponent makes a mistake, sometimes it’s hard to re-orient your thinking to the fact that you might actually win. When he played his move, he probably thought, “Maybe this will worry my opponent enough that she will take a draw.” (Both of them had less than 2 minutes left.) Chu was worried, all right — she saw that she was dead lost, so she had to take the draw offer.

The shocking turn of events must have left both players shaking their heads, but the bigger picture was that Armenia was still alive, needing only to win one game to catch up. And Karen Grigoryan came through for them, winning a pawn and outplaying Xu Xiangyu in the endgame. Now the match was even at 7-7, with the games Zaven Andriasyan – Wang Yue and Samvel Ter-Sahakyan – Ni Hua still going.

But not for long! Both of those games were quickly agreed to draws, as both teams apparently preferred to play it safe and go to the tiebreaker instead of attempting to win in “regulation.”

Here are how the tiebreakers work in the PRO Chess League. It starts with #4 playing #4. Then when one of them loses, he is eliminated and the #3 player for that team replaces him for the next game. And for each game after that, the loser gets eliminated and the next higher player on that team plays next. If there is a draw, then both players get eliminated. Finally, if it comes down to #1 versus #1, they keep playing games until someone wins.

The thing I hate about this format is that 90 percent of the time it will come down to #1 versus #1, and the earlier games are essentially meaningless. It turns a team competition into an individual competition — who has the best first board? I hope that they’ll look for a better tiebreak method in the future, but I don’t think it’s very likely because it seems to be popular with the fans (all of the fans except me).

So, indeed, the match did come down to #1 versus #1. For historical completeness, I will report that China’s #4 beat Armenia’s #4, then Armenia’s #3 beat China’s #4. Then the two #3 players drew, eliminating each other. Then the two #2 players drew, eliminating each other. Thus, the whole season came down to Wang Yue versus Zaven Andriasyan.

Although Yue is rated much higher than Andriasyan in  regular (long time control) chess, Andriasyan is a beast at blitz. He had won a “Titled Tuesday” tournament at earlier this year, against a number of super-GM’s, and I think that this was the basis of Hikaru’s forecast that Armenia would win the tiebreaks. The first game of the tiebreak was very exciting, as Yue defended brilliantly against Andriasyan’s attack and even seemed to have some winning chances in the endgame, but they ended up drawing. The second game of the tiebreak was a quick and relatively uneventful draw. But in the third game, Andriasyan’s pressure was too much. In a Q+B versus Q+B endgame he had an extra pawn and a much more active queen, and he ended up skewering Yue’s king and bishop. Then, for good measure (already a piece ahead) he skewered Yue’s king and queen! At that point Yue threw in the towel, and Armenia celebrated their hard-earned victory.

The number of viewers online got as high as 27,000. Does anyone know how this compares with the number of online viewers of “traditional” events like the world championship or the Olympiad? All I can say is that for a chess event, that sounds like a lot of viewers to me.

Congratulations to commissioner Greg Shahade and the whole team. The chess was far from perfect, but if you want perfect chess, go watch AlphaZero beat Stockfish. This was an exciting, fun, and somewhat chess-like competition, and I think that is the main point.

Print Friendly

{ 1 comment }

Live PRO Chess League Playoffs!

April 7, 2018

Today was a very exciting day in the history of the PRO Chess League: the day of the first in-person playoffs. Four teams — the Chengdu Pandas, Ljubljana Turtles, Armenia Eagles, and St. Louis Arch Bishops (the defending champions) came to San Francisco to face off in the two semifinal matches. The winners will go […]

Print Friendly
Read the full article →

Most Popular Posts (2018 Edition)

April 3, 2018

About once a year I take a look at my site statistics to see what my most popular posts were. Then for the rest of the year, I ignore what’s popular and just write what I feel like writing! I do think that new or infrequent visitors might appreciate a guide to what some of […]

Print Friendly
Read the full article →

My Game with Magnus

April 1, 2018

Happy April Fool’s Day! I really am going to show you my game with Magnus, only it isn’t the Magnus you’re thinking about. In 1995 I played in the World Open, and for once it kind of lived up to its name. In round one, I played against Leonid Filatov, a U.S. player born in […]

Print Friendly
Read the full article →

(Off Topic) Absurd and Adsurd

March 30, 2018

With apologies, I am going to go off topic today and talk about math. Some chess players like math, so I hope you’ll be interested in this post! Big news! Today, for the first time in 23 years, I’ve had an article published in a mathematical research journal. If you want to read it, go […]

Print Friendly
Read the full article →

Why We’re Jubilant

March 28, 2018

This is a Facebook post I wrote today, and I decided to post it here too. (It’s very short!) One of my Facebook friends, Mike Zaloznyy, started a thread asking why there was so much rejoicing over Fabiano Caruana’s qualifying to play a match for the chess world championship with Magnus Carlsen. In particular, Mike […]

Print Friendly
Read the full article →

It’s Official! American Challenger for World Championship

March 27, 2018

Congratulations to Fabiano Caruana, who won his game against Alexander Grischuk today and became the first U.S.-born challenger for the World Championship since Bobby Fischer in 1972. (*) For any readers who don’t know, Caruana was born in Miami and lived in the U.S. for his pre-teen years, but when it became clear that he […]

Print Friendly
Read the full article →

Thirteen is lucky for Caruana!

March 26, 2018

So far I haven’t written anything in my blog about the ongoing Candidates Tournament to pick the next challenger to Magnus Carlsen. After all, what can I tell you that you can’t read on a thousand other websites? However, today’s round was too exciting not to write about. It was the 13th (second-to-last) round of […]

Print Friendly
Read the full article →

Initiative, Initiative, Initiative

March 21, 2018

The three most important things in playing winning chess are initiative, initiative, and initiative! That has been my philosophy for many years. Although there are some exceptions and some other ways to play, I still believe that many chess games boil down to who can seize and keep the initiative. Three weeks ago, in the […]

Print Friendly
Read the full article →