Friday, April 27, 2007

Counting Error

Here's an example of a counting error from one of my Blitz games. It's not very hard at all, but with the clock ticking I wasn't able to calculate it correctly.

Analyze the consequences of 1.Ba3.

1.Ba3 Bxc4 2.Bxd6 Bxd3 3.Bxf8 Bxf1 and Black is a piece ahead. Best then is the desparado 4.Bxg7.

I looked at this position two different ways afterwards. First I did "He takes, I take..." and tried to keep a running tally. I could see that Black would be a piece ahead, but I had to concentrate. One contributing factor was that taking "snapshots" at certain points before quiescence was misleading. Sort of like:

1...Bxc4 2.Bxd6 ("equal so far") 2...Bxd3 3.Bxf8 ("I'm up the exchange") 3...Bxf1 4. Bb3 or 4.Kxf1 ("I return the exchange") to give the illusion that everything would be equal, rather than that White would be a piece down.

However, if you just barrel down the main line and add up pieces, you can see: "In the next three moves, Black takes N, B, and R and White takes B and R. At that point, whether both bishops save themselves or both bishops are captured, Black is still up a piece."

So it may be that it's easier to calculate "I take this and this, he takes that and that and that", rather than "I take, he takes, I take...". I'll have to try this out in practice. This method may make you more likely to mis-evaluate a side variation (e.g. 1...Bxa3 or 2.Bxc4) since you're just gunning down what you perceive the main line to be, but if nothing else it can help you "check your math".

Friday, April 20, 2007

"Hall of Fame" Databases

No sooner do I post my intent to not clutter my blog with my own ICC blitz games, than I play another game that's actually demonstrates my previous advice.

I'm testing out the Chess Publisher web tool for posting entire games. Here's why I'm posting this game:

1. I've said that you can use a chess engine such as Fritz to "mine" your blitz games for tactics, but emphasized the "Hall of Shame" databases of your errors. "Hall of Fame" databases, where you successfully executed a tactic, are also a good idea, and a morale booster as well. Sometimes, after focusing on all your shortcomings, it's nice to go over examples where you did play the best moves. This game demonstrates the theme of the back rank weakness. It's a cute example, because there were two separate weaknesses on d8 and e8 that were exploited. You can find similar examples in tactics books, but when the position occurs in your own game I think it has greater impact.

I've been a bit lazy lately, but I'm trying to routinely dissect my blitz games and save the tactics in their own files. I separate them by "tactics I successfully executed" and "errors where I missed a tactic", and I further separate them into 1, 2, 3, and >3-move tactics (grabbing an en prise piece, or leaving a piece en prise would go into the 1-move folder; a knight fork would go into the 2-move, and a "Chernev and Reinfeld: How to See Three Moves Ahead"-style tactic would go into the 3-move). When they get sizeable enough I'll have my own databases of real-world, simple tactics that I can use for drills.

2. The game demonstrated how understanding is more valuable than opening preparation. I was actually out of book on move 4, more or less. Even though I've played the French as Black for over a decade, I rarely face the Tarrasch, and I've only recently started playing it as White. I'm sure I've had the position before, but never bother memorizing the continuation. However, I have a decent understanding of how to play against an IQP (Isolated Queen's Pawn) and played according to the position. A quick check with Fritz and Chessbase shows that I actually played the opening quite credibly. I knew that I was supposed to play Bb5 at some point, but since I don't really understand why (yet...I'm going to check out this move order when I have time)I played my own moves and still got a good middlegame.

This game was a really, really rare example of error-free play on my part (error-free meaning my move never differed from Fritz's best move at about 10 ply by more than about 0.3 pawns).

My opponent was rated in the mid-1400s at blitz, about 80 points higher than me.

1. e4 e6 2. d4 d5 3. Nd2 c5 4. Ngf3 Nc6 5. exd5 exd5 6. dxc5 Bxc5 7. Nb3 b6 8. Be2 Be6 9. c3 Nge7 10. O-O O-O 11. Bg5 Qc7 12. Bxe7 Nxe7 13. Nbd4 Rad8 14. Re1 Nc6 15. Qd2 Nxd4 16. Nxd4 Kh8 17. Rad1 Be7 18. Bf3 Rd6? 19. Nb5! Qb8 20. Nxd6 Bxd6 .

A point-count comparison indicates that White is only up 1.25 pawns
at this point (the exchange is worth 1.75 pawns, but Black's bishop pair is
worth 0.5 pawn). However, white's pieces are actively placed, whereas Black's
are defensively placed. Fritz's evaluation of about +2.9 pawns for White is
probably more accurate. Fritz probably sees that the d pawn is indefensible
in the long run. However, with two bishops pointed at the kingside, White has to be careful about a kingside attack, especially after the next move (I'm always worried about moves like...Bxh3). 21. h3 Bf4 22. Qd3 Rd8 23. Qb5! Setting up the following tactical opportunity...Qd6?

24. Bxd5! Bxd5? 25. Rxd5! Qf6 26. Re8+ (26.Qe8+ and 26.Rxd8+ also lead to mate.) 1-0.

Note that, on Black's part, this is an example of both a counting error and a quiescence error. It seems like Black has just enough defenders of the d-pawn, but if Black were to capture three times on d5 the position is not yet quiet because Re8 would mate.

Wednesday, April 18, 2007

Applying Endgame Knowledge

I don't intend to have a lot of my own games on this site. My intent is to keep this site focused on material that will help other people as well as myself. It would be easy to list a bunch of ICC blitz games that I found interesting, but few others would.

However, if it demonstrates a general point that I want to reinforce, I'll succumb. Here is an example of how a little bit of endgame knowledge goes a long way.

I've known for a long, long time how to play Queen vs. Pawn on the 7th rank positions, but I've always been a bit sloppy in the execution (not "crisp" as NM Alex Betaneli would say). This endgame was very well covered in Silman's Complete Endgame Course, and even though I have only had this book about a month, I have put this refresher course to practical use and executed this endgame in several online games. The following endgame still isn't "crisp", but it possesses much higher turgor pressure than my old efforts.

53. Kc2?

53. Ke2 draws: 53...Kc3 (... Ke5 54. Ke3=) 54. Ke3=. After White's actual move it was easy to calculate the pawn race and see that I would win provided I knew how to win the Q vs. P endgame.

53... Ke3! 54. Kc3 Kf4 55. Kc4 Kxg4 56. Kxc5 Kf5 57. d4 g4 58. d5 g3 59. d6
g2 60. d7 g1=Q+

This is the position I calculated up to before playing my 53rd move. It was important to visualize that Black queens with check. What follows is not the quickest, most crisp win, but I used the basic technique and got the job done: 61. Kc6 Qc1+ 62. Kd6 Qd2+ 63. Ke7 Qe3+ 64. Kf8 Qd4 65. Ke7 Qe5+
66. Kf8 Qd6+ (66... Ke6 !? is a cute move Fritz found, but there's no need to
take such risks in an actual game. The idea is 67. d8=Q {67. d8=N+ is a better practical shot but it's still mate in four} 67... Qh8#) 67. Ke8 Qe6+ 68. Kd8 Kf6 69. Kc8
Qc4+ 70. Kd8 Ke6 71. Ke8 Qc6 0-1

Because I was familiar with the basic Q vs. P endgame, I was able to calculate ahead 15 ply to the queen check and evaluate it as a won endgame for Black. Even though I absolutely suck at blitz (you should have seen the rest of the game! oy...) I was able to do this in a few seconds of counting.

One of the values of studying endgames is that it allows you to spot transitions into simpler, winning endgames from more complicated ones. This is especially important if you are deciding to enter a pawn endgame, because of their concrete and unforgiving nature. For example, after you master K+P vs. K, you can calculate an endgame with an outside passed pawn down to these K+P vs. K endgames. Once you understand how to win an "outside passer" endgame, you can spot an opportunity to simplify from a B vs. N endgame to a won "outside-passer" pawn endgame. And so on.

Finally: although I won the endgame, there were numerous, simple tactical errors throughout the game, which demonstrates the supremacy of studying tactics and correcting sloppy thinking. If I had played better earlier on, I would not have had to worry about the endgame. However, endgame knowledge is immensely useful as a "safety net", especially since a lot of people hate studying endgames. Also, some opponents, even stronger ones, will try to simplify down to an endgame if they feel threatened. If your endgame skills are better than theirs, they may be in for an ugly surprise.

Heisman's Latest Article: Counting Errors Revisited

Dan Heisman's latest article revisits counting errors ("I take, he takes, I take...oops.") with some more elaborate examples. None of the examples are of mind-blowing complexity, but for several I had to sit and think. For example:

The moves played in this game were (after 1...Nbd5) 2.Ne5? Nxf4 3.Nxf6? Bxf6?? 4.Be4?? Ng6. In particular, Black's third move looks natural, but is not best. If Black's correct 3rd move isn't immediately obvious to you as you play through this example, go straight to Heisman's article, do not pass go, and do not collect 200 ratings points (at least, not until you're done reading through the Novice Nook archives there).

If you routinely mine your games (including Blitz) for tactical errors, you'll be finding plenty of these errors in your own games and can include them in your "Hall of Shame" files.

Wednesday, April 11, 2007

When Pawns Lie

I've been slowly reading through Chess Self-Improvement, by Zenon Franco. This is actually a fairly advanced quiz book, so I don't dwell too long on the questions. I'm reading it more as a collection of annotated games.

I was struck by the game Byrne-Vaganian, Moscow 1975. Look at this position:

Now, someone like me would look at this position and invoke the teachings of Silman: in closed positions, you want to play on the side that your pawns point towards. This is because this is the region of the board where you have a space advantage. So, it's pretty clear that in this game White is going to attack on the kingside, and Black is going to try and attack White's weak pawns on the queenside. Right? I mean, that's what the pawn structure is telling us.

However, after White played 16.Bxg6, Black responded 16...fxg6!, and doubled rooks on the half-open f-file. White later had his queen on the half-open b-file (trying to penetrate into d6 via b4) but to no avail. Black won this encounter through kingside play.

To me, it seems there were a couple key reasons for this violation. First, each side had half-open files on the sides where they didn't have their space advantage. Second, the weak white pawns on the queenside ended up being a diversion. Black could still keep a queen and bishop eying the queenside while he concentrated on the f-file invasion, and White's queen got stuck on the queenside trying to either hold the pieces together, threaten the b-pawn, or invade at d6.

This shows that rules have exceptions, and that control of a half-open file may compensate for a spacial disadvantage in that sector.

Wagons, East!

If I don't post as much over the next month or so, there's a good reason: it appears that my wife and I will be moving to the East Coast! And rapidly, too....we're heading out to look for houses this weekend.

The good news is that I should be moving to an area with a lot more chess activity than where I live now. So, instead of getting only 1-2 tournaments a year in, I can hope for some regular OTB experience.

I'm in the middle of drafting a self-improvement plan for myself, but I'm not going to embark on it until I'm settled at my new digs. I don't think it'll be as intensive as a de la Maza-esque program, but something more structured than what I'm doing now. I'm going to try taking my own advice and seeing if I can improve my strength.

Wish us luck! I'll still be checking the blog, but probably won't be writing novel-length posts for a bit.