25 July 2010

Symmetry Misjudged

In Castling Misjudged, I showed how an error in judgement led to a loss in one game of SchemingMind.com's 2009 Chess960 Dropout Tournament. The other game I lost, which was enough to eliminate me from the event, was also due to an error in judgement, although of a different kind.

Note the symmetry in the diagrammed position. It was even more striking before Black's last move, 16...Rcd8, when the move 16...Bb7 would have achieved perfect symmetry.


After 16...Rc8-d8

The primary tactical consideration for both sides is the sequence starting Nxd5 (or ...Nxd4 for Black), followed by advancing the e-Pawn two squares after the capture. I decided that 17.Nxd5 was premature and played 17.Rfd1 instead, judging that 17...Nxd4 was also premature. My opponent decided otherwise, played that move, and ended up with a long-lasting initiative that eventually netted a Pawn. Here's the full game score, courtesy SchemingMind.com.

[Event "2009 Chess960 Dropout Tournament, Round 5"]
[Site "SchemingMind.com"]
[Date "2010.03.06"]
[Round "5"]
[White "bemweeks"]
[Black "wilfried"]
[Result "0-1"]
[Variant "fischerandom"]
[SetUp "1"]
[FEN "brnqnkrb/pppppppp/8/8/8/8/PPPPPPPP/BRNQNKRB w KQkq - 0 1"]

1.d4 d5 2.Ncd3 O-O 3.Nf3 b6 4.b3 Ncd6 5.O-O g6 6.g3 Ne4 7.Nfe5 N8d6 8.f3 Ng5 9.Bg2 Ne6 10.Qd2 f6 11.Ng4 Bg7 12.Bh3 Ng5 13.Bg2 Qd7 14.Rbc1 Nf5 15.Bb2 Rbc8 16.Nf4 Rcd8 17.Rfd1 Nxd4 18.Bxd4 e5 19.h4 Nf7 20.h5 g5 21.Bh3 Qd6 22.Ne3 Nh6 23.Ne6 exd4 24.Qxd4 Rde8 25.Nxf8 Qxg3+ 26.Bg2 f5 27.Qd3 Bxf8 28.Nxf5 Nxf5 29.Qxf5 Rxe2 30.Qh3 Qf4 31.Qg4 Qe3+ 32.Kh1 Bd6 33.h6 d4 34.Qc8+ Kf7 35.Qf5+ Ke8 36.Qd3 Qxd3 37.Rxd3 Bf4 38.Rcd1 Be3 39.Rxd4 Bxd4 40.Rxd4 Ke7 41.Rc4 c5 42.a3 Bd5 43.Rc3 Re6 44.Kg1 Rxh6 45.f4 Bxg2 46.Kxg2 g4 47.a4 Re6 48.Kg3 h5 49.a5 Re2 0-1

How could the better developed side end up tactically worse in a symmetrical position? That is a question for which I have no answer.


HarryO said...

Wow, what an extraordinary game! Firstly, if you did not use computers to aid you, both you and your opponent are without question elite players. There were no serious midgame blunders in the entire game and white seems to pay the penalty for pushing the h-pawn.

I have had to use Rybka3-960 to help me sort through this incredible game. My initial reaction is that from the super symmetrical moment, it is a constant race to maintain the initiative by tactical motif as you have also seen. The point of divergence for me seems to be 19. h4?!

At that point white gives black another tactical motif (the weak g3 square and king) while at the same time removing a tactical motif for himself (better? is Bxe5..fxe5 and now the Ng5 square is weak). So it appears black develops stronger realisable initiatives against the white king.

Did you see the nice move Rybka960 finds?:
22 Ne6!? exd4 (only move)
23.Kh2 c5
24 Nxd8 Rxd8

At this point, white is up in material, his king more secure and as far as I can tell is also better positionally by a small margin. Black's light bishop is blocked up and his dark bishop is only marginally better but with no targets. Black's queenside push looks nice but how strong is it really?? Rybka thinks black has the edge!? I wonder how Kramnik would play here as white? It is possibly too much for a computer to understand?

It is very complex, almost stellar in possibilities. A stellar game, thanks for sharing that with us.


Mark Weeks said...

Thanks for your comments, HarryO. The site SchemingMind.com is for correspondence chess, which explains the relatively error free play. It is rare for competent correspondence players to blunder, with or without computers. The errors are almost always more subtle, like misjudging long term factors. I'm certainly not an elite player (my peak FIDE rating was 2250), but my opponent, who was rated about 200 points higher than me, might very well be. It's hard to say exactly because the site permits the use of pseudonyms, meaning you don't really know with whom you are playing.

I have a copy of Rybka, but it's a version from before chess960 was added. I don't often use an engine to analyze chess960 openings. My experience from studying the CCRL games is that the engines don't play the opening particularly well. It's better to apply logic. I often choose an opening move that the engines haven't tried and so far I have had good results doing that.

As for your specific suggestions, I'll take a close look at them when I get a chance. I record most of my ideas in an ongoing PGN record of the game and can use that as a reference. - Mark

HarryO said...

Wow Mark, it all boils down to the short term sacrifise 20.Ng6!

I realised I didn't answer your question properly. You are asking why is it that from a super symmetrical moment with white better developed, black can play Nxd4!? In a wonderful twist of logic, it is white who breaks the symmetry prematurely not black that allows Nxd4.

It comes down to the placement of the rooks on white's back rank and what those rooks are doing both in defense and offense. In the real game, black expends tempo playing his rook from b8 to c8 to d8 yet this does not hurt him! The critical moment I think is at move 17:

17. Rfd1 is what white played. It truly breaks the symmetry. You will notice that not only are the bishops displaced, but so are the rooks. In this situation, black may take with Nxd4 and is ok.


17. Rcd1 is what you should have played maintaining an almost perfect symmetry. In this situation the symmetry of the rooks is perfect but black may NOT take with Nxd4. It is only white's bishop on b2 that disturbs the super symmetry, but it appears that is not relevant to the main line.

If we now play through the main line with what you played and what the proposal play is, we discover why black can or cannot take with Nxd4.

17) Rfd1? (breaking the symmetry)... Nxd4!?
18) Bxd4 ... e5 (the immediate point)
19) Bxe5!... fxe5
20) Nxd5 ... (Nxg6? ... Nxf3+!!)
20) ... Qxd5
21) Qxg5 ... Qxd1+ (not clear what Qxg5 achieves)
22) Rxd1 ... Rxd1+
23) Bf1 ...

Here we have a truly bizarre situation that is beyond me. White has N+Q+P v 2R + B! It looks dynamically balanced to me. Let's say it's equal!? In other words, the actual real game move 17 produces some kind of equality I do not understand.

Now if you were to have maintained the symmetry instead of breaking it, now white would have real advantage that is much easier to see:

17) Rcd1 (maintaining the symmetry) ... Nxd4?
18) Bxd4 ... e5
19) Bxe5!... fxe5
20) Nxg6!! (possible in this version ... Nxf3+??)
20) ... hxg6
21) Qxg5 (now it is clear what Qxg5 does!)
22) ...Qd6
23) e4 ... c6
24) f4 ... exf4
25) gxf4 ...

We see why it was so valuable to leave the white rook at f1. It prevents ...Nxf3+! By white timing the break in symmetry to perfection, he now has a huge king attack involving the queen, knight, f-pawn and the f1 rook. In some lines, the f1 rook also can play a defensive role by playing Rf2 to guard against check and still maintain access to the f-file and the f3 square.

So this example seems to be an absolute cracker example of the effect of symmetry for both sides and how and when symmetry should be broken for maximum effect. As a very tentative general rule of thumb "do not break the symmetry if you have the move!?!?!"

To be a perfect player, you would have needed to see that the d-pawn is poison for black at move 17, but only if the symmetry is maintained that little bit longer until the decisive break 20) Ng6. The pawn is poison because with a rook on f1, white has a big king side attack.

However there is a simpler lesson for mere mortal chess players. If deciding on moving a rook that is on the back rank, do not move a rook if that rook will end up with less potential than before. The rook on f1 does at least three things. It guards with Rf2 and defends f3 and it backs up a king attack. It does not do as much on c1.

Amazing stuff. Such tiny changes with such huge outcomes. The question is, would standard chess be able to throw up an example quite like this? Possibly, but my suspicion is that this example is just the tip of the Chess960 iceberg.


Mark Weeks said...

Yes, I agree that 17.Rfd1 was a mistake. My plan was to prepare c2-c4, but I was never able to execute it. I'm not sure what I should have played instead. The position could use deeper analysis, but I have other fish to fry for now. I just started round 3 of Schemingmind's 2010 chess960 dropout tournament and, unlike traditional chess, the early moves of a chess960 game require a lot of attention. Maybe later... - Mark