Sunday, March 18, 2007

Patterns of Error

diagram 1 White to play and win.

diagram 2 Black to play and win.

I have recently been coaching a young player, which I am learning is a little different from teaching chess to kids. For one thing, you can go much deeper into opening lines, middlegame themes, and endgame problems with an interested individual than you can with a mixed group of antsy boys. You also get to tailor your lessons to the specific needs of your student, which you can identify by analyzing his games.

Coaching a young chessplayer reminds me of tutoring writing (at which I have much more experience), and some similar issues arise, especially as regards "patterns of error."

When tutoring writing, we speak of "patterns of error," or the specific mistakes in grammar, syntax, spelling, etc. that a student makes in his or her essays. Identifying and working on just a few patterns of error at a time keeps students from getting overwhelmed and helps them learn to edit their own work. Too often, writing professors (especially those with a background in editing) will mark every error they see on an assignment, leaving students feeling overwhelmed by the task of making corrections. The tutor can perform a valuable act of triage by classifying errors, prioritizing them (from "fatal errors," such as subject-verb agreement, to trivial ones, such as occasional dropped articles in the prose of a non-native speaker), and then showing the student how to correct the most important ones. You can also sometimes help in getting to the root of the problem.

For example, I've seen students who present with problems in subject-verb agreement ("Gould say [sic] that evolution is a misuse [sic] term"), errors in number ("Wooly mammoth [sic] are extinct"), and errors in tense ("They use [sic] to think fossils were evidence that dragons once existed"). The student could work on those errors with the help of a grammar text, where he'd find them listed under subject-verb agreement, constructing plurals, and using the past tense. But a good tutor might notice that the larger pattern of error really boils down to issues with sub-vocalizing -s and -ed endings. Therefore, the best way to address these errors is for the student to pay attention to the words where he is not hearing the endings. He should collect a list of these words and memorize them. If the student still cannot hear the proper endings, at least he will see the problem words and learn to make the correction.

It seems to me that patterns of error in chess are quite parallel to those we see in writing. My chess student, for example, has been making some critical errors that involve not recognizing his opponent's threats. Of course, this is a fairly common error type in developing players (some might say in all players). But then I noticed a deeper pattern emerge: almost all of his errors involved threats on dark-square diagonals. It's almost as though he's especially blind to attacks by a dark-squared Bishop, as the following two diagrams (both from his recent games) illustrate.


diagram 3 White has just played 14.f4? to win the Bishop
at d3, which allows Black to win by force.


diagram 4 Black overlooked White's
threat and played 16...dxe4?

Fortunately his opponents must suffer from the same color-blindness, because they both overlooked the threats as well! He can't rely on such lucky breaks in the future, especially as he begins to play stronger opposition. In the first example, 14...Neg4! would have been absolutely crushing, due to the deadly threat of ...Bc5+. And in the second example, White should simply win the Exchange with 17.Ba5!

After I noticed this dark square problem, I remembered other examples from his play. The first game we ever played together, for instance, went 1.e4 e5 2.Nc3 Nc6 3.f4 Bb4!? 4.fxe4?? Bxc3 5.dxc3 Qh4+ etc. And the other day we were discussing the Queen's Gambit and he asked me why after 1.d4 d5 2.c4 e6 3.Nc3 Nf6 4.Bg5 Nbd7 5.cxd5 exd5 White did not simply play 6.Nxd5?? I was surprised he would make these errors since I'd rate him around 1600, yet any 1600 player would see that these are absolute blunders. Clearly it is a pattern of error and I think it is one he can train himself to control.

Classifying his error is an important first step. Next we will have to work through some problems with dark-square attacks or dark-square threats so that he begins to train his recognition of those patterns. We could begin with threats on the dark square diagonals around the King (perhaps looking at 1.d4 Nf6 2.Nd2?! e5!? 3.dxe5 Ng4 4.h3?? Ne3! as in the supposed shortest master game or 1. d4 Nf6 2. Bg5 c5 3. c4 cxd4 4. Nf3 e5 5. Nbd2 Nc6 6. Ne4?? Nxe4! as in a nice miniature by Scott Massey). Then we will work on other situations with dark-square attacks.

Once you notice a single pattern of error, others begin to emerge. One interesting pattern of error can be seen in the first two diagrams at the beginning of this article. In both these games, from the same tournament, my student accepted a draw against opponents rated over 1600. In fact, in the second diagram his opponent offered him a draw in the diagram position itself and, without thinking (despite plenty of time on his clock), my student took it. Yet in both positions, as you have already likely observed, he could have won by exploiting a pin. In the first position he should double Rooks on the b-file with 1.Rb2! followed by Reb1, winning the Knight at b5 which is pinned to the Rook at b8. In the second diagram he could win a piece by 1....Re1! followed by 2...Ne3, and if 2.Nc4 Ne3! still wins the Bishop after 3.Nxe3? dxe3 with the deadly threat of 4....e2 etc. In both cases, he simply did not recognize the possibilities of the Rook pin, so he ended up with just two draws out of two games instead of the two wins that he deserved. The parallels between these two cases are uncanny, yet rather typical of patterns of error.

We cannot recognize what we have never seen before, and we will never recognize it until we've consciously trained ourselves to see it in the first place.

You can track patterns of error on your own by putting together a database of your games and using Fritz or another strong program to blunder check. Then use the computer to create diagrams of situations where you've made mistakes (you can simply use Edit>"Copy Position" and paste into Microsoft Word--if you have the right chess fonts installed). Once you've collected enough diagrams, look for patterns of error. Inevitably something will begin to emerge. Of course, a chess coach can be especially helpful in designing a training plan to address that error in your play. But just becoming aware of the pattern or patterns is a good first step.


looking for troubleThere have not been many books that can help students train themselves to recognize their patterns of error and correct them. Among the best of those few is Dan Heisman's Looking for Trouble: Recognizing and Meeting Threats in Chess. You can read an excerpt of the book at his "Novice Nook" column at ChessCafe. I am a big fan of Heisman's work and especially liked his piece on "Quiescence Errors," which discusses cases where players end their analysis too soon, usually stopping on some stereotyped move ("Oh, if he does that he loses his Queen, so that can't be good...") rather than looking a little further to recognize that they've really blundered ("...but wait! I win his Queen but I also get mated! Yikes!") That may be one of my own patterns of error, though I tend to make it in a more positive way, where I might not explore a line sufficiently because it involves an initial sacrifice yet, if examined far enough, could lead to a winning attack.

Heisman has written a lot on the issue of error for beginning chess players. The better pieces among his "Novice Nook" columns on the topic include "The Seeds of Tactical Destruction," "A Counting Primer," "Revisiting the Seeds of Tactical Destruction," "The Most Common OTB Mistakes," "Is there a Win?" and "Is It Safe?"

danger in chessAnother interesting book on the theme of error is Amatzia Avni's Danger in Chess: How to Avoid Making Blunders. I'm not sure it will be as helpful to the average player as Heisman's work, but I enjoyed looking at its examples. Avni's book may also come the closest I've seen to offering what I'd call a "Grammar of Chess Mistakes," which can be a helpful thing to have for categorizing your errors. Avni's categories include "leaving the king with insufficient support," "weakness of the eighth rank," "entering a lasting pin," "capturing poisoned pawns," and "placing pieces without escape routes." Most of his examples (all drawn from GM games) are more psychological than tactical, however, and amount to three problem areas: (1) underestimating your opponent, (2) underestimating the situation, and (3) ignoring your opponent's body language (as when he appears a bit too happy to enter a line you thought was bad for him). He then goes on to talk about ways players can manipulate these psychological effects to their advantage to misguide their opponents. I am not sure how useful this is to developing players, but his examples concerning "The Art of Deception" are the most interesting in the book. It might be an interesting project to collect games where the psychological element can be seen in the moves themselves. My favorite example of that is Kupchik - Capablanca, Lake Hopatcong 1926, where Capa appears to weaken his kingside in order to lure Kupchik into an attack on that flank, which only serves to make his own attack on the queenside all the more effective. A similar case might be Geller-Euwe, Zurich 1953.

Though most of his examples seem rather esoteric, Avni does offer some useful ideas for students and coaches. Basically, if you want to avoid error, you need to train yourself to do the following:

  • Begin your thought process by looking for your opponent's threats rather than focusing on your own attacking ideas.
  • Maintain a self-critical attitude and always double-check your calculations.
  • Check actively for possible dangers and try to "think for the other guy" so that you begin to anticipate his ideas.

I especially like his suggestion that students study double-edged games so that they are always studying tactics with their opponent's threats in mind.

One of Avni's better examples comes from the game Kagan-Kaldor, Israel 1971, which illustrates his three principles quite well:


diagram 7 Kagan - Kaldor, Israel 1971
Black to play (the board is from his perspective).
What dangers lurk?

Black to play (the board is presented from his perspective) continued 40...Re2+? 41.Kxe2 a7 and the pawn cannot be stopped. No doubt Kaldor thought he was winning. But Kagan was able to force a draw by 42.Rh6+ Ke5 43.Rh5+ and Black dare not play 43...Kd4?? because of 44.Kd2 followed by Rd5#. One might also note that Black cannot escape in the other direction: after 42.Rh6+ both 42...Kf7 and 42...Kd7 allow 43.Rh7+ followed by 44.Ra7 which stops the passed pawn and wins.

Kaldor did not double-check his calculations, and he did not see what his opponent might be able to do. All he saw was the most forcing way to get the pawn to the queening square.

Controlling most patterns of error comes down to double-checking your calculations from your opponent's point of view. This lesson was brought home to me recently as I was looking at the game Tarrasch-Mieses, Goteborg 1920 in Irving Chernev's Logical Chess, Move by Move.


diagram 8 White to play and not blunder....

The diagram above appears on page 202 of the book, and I remember thinking immediately on seeing it that 25.b5 wins. Then I looked at Chernev's note: if 25.b5? Na5+ 26.Kb4? Nd5+ 27.Kxa5?? Ra8#! I could easily have walked into that trap, since the moves leading up to mate seem so natural from White's perspective. Once you start to think from the other side of the board, though, the mating trap becomes obvious. Fortunately, Tarrasch was aware of the danger and checked his calculations, going on to produce one of the most error-free games of his career.

It takes training to weed out our errors. But if you develop the right habits of thought, you will keep errors under control and increase your winning percentage significantly. I look forward to seeing my student do just that.

6 comments:

Blue Devil Knight said...

Good stuff. I think I need a new coach!

Tom Chivers said...

Very interesting post. The double edged game part indeed is a good idea I'd not thought of before (and probably should do myself.)

I realised when I came back to chess a year and a half ago that I frequently overlooked even elementary piece-trapping tactics - both OTB, correspondence, and the CTS I was using to get me back in the OTB mood. I made a conscious effort when ever I smelt a rat to double, treble check, and think I have by and large corrected this.

I also have a friend I play correspondence chess, who out of the 10 games we've played - I've won all ten of them with knight forks. In some cases the forks finished off a won game; in others they shifted the result to me. I told him after the first four he has a blind-spot for knight-forks and ought do what I need for piece-trapping tactics but for knight-forks, but evidently he failed to follow my advice.

Heh ho.

Anonymous said...

Another great post, Michael. But isn't this all just common sense? We all have weaknesses and strengths, and learn different things at different times, whether it is writing or chess. I'm not disparaging your post by saying this is common sense; the argument is elegantly presented. BTW, here's where patterns of error in chess meet writing: open any chess book these days, and the punctuation and grammar is poor.

Howard Goldowsky

Michael Goeller said...

Thanks for the feedback, Howard. But as Tom Chivers remarks, it may all be "common sense" (as you say) but many players persist in their patterns of error until they take conscious steps to remedy them, as Tom has. The same goes with writing: unless a writer works at correcting his or her patterns of error, those errors will persist. A good tutor or coach can make a significant difference in how quickly those errors get identified and controlled. That may be common sense, but it is not common practice.

Meanwhile, what you say about the generally poor writing in chess books is quite true--and generally results from GMs and IMs with great chess knowledge writing books in English (the most lucrative language for publishing) when they are neither native speakers of the language nor professional writers. What's more, the editing standards in chess are relatively low: there is just not enough money in the venture to hire enough good editors and the market is obviously willing to accept less than elegant prose.

Anonymous said...

A highly original piece. Don't see what that guy means by common sense. That comment has nothing to do with what you wrote.

Ryan Emmett said...

Terrific post! Very interesting indeed.

I'd like to say that I find your blog excellent and enjoy reading it very much! Keep it up!