Saturday, February 21, 2009

Interesting criticisms of the Circles

The following is a response I emailed to a reader generous enough to send me some of his well-considered criticisms of the Circles method of tactical training (for those who don't know this method involves doing the same set of problems many times until you can do them really fast--the first time through you spend ~10 minutes/problem, and each cycle you solve them faster until you can do them really quickly without thinking).

I don't claim to have any final answers here. He brings up enough interesting, and complicated, issues that I am surely wrong on some of my points. I thought the points he made were interesting enough to merit putting them out here in the improvement blogosphere.

His concerns are in italics, my responses are in normal font.

Somewhere I read a paper that says it's better to see the same material presented in several different modalities, than to repeat the same modality. The latter helps retention, but the former results in stronger transfer. (But I cannot remember the title and authors. I should have read it aloud and traced it on my palms!) This suggests that it is better to do 7000 unique problems than Seven Circles of the same 1000.

It seems this is orthogonal to the Circles question. Different sensory modalities isn't the same as different problem sets. However, as a separate topic you probably want to see many examples of a particular tactical motif in the problem set. I discussed this a bit here in my most considered defense of the Circles against a common criticism (see the comments there too as I expand on the ideas in the main entry).

Separately, we know that there are hundreds of thousands or millions of meaningfully distinct chess patterns (after all, GMs are said to have a grasp of 100,000). This is another reason to prefer the 7000.

Yes, but 1000 is a good start, and I don't think you are narrowly memorizing just those 1000 problems as I argued in my post linked above. Also, if your goal is to learn the problems cold so you can do them effortlessly, then 7000 seems a big chunk to bite off all at once.

The spacing of repetitions should be close together at first, then further apart. Seven Circles has it backwards.

I have seen before that if your goal is to remember X, you should review it within 24 hours, so I think you have an excellent point. However, it depends on your goal. Many (such as MDLM) look at first few circles as primarily being for calculation training, not as something optimized for memory construction. However, to form memories, perhaps a modified circles program would have you do 50 a day, really quickly, and review them each day.

And besides all that, chess skill goes beyond mere memory retention, and includes a lot of uncharted higher level processes (doesn't it?). On that basis we shouldn't we be doing something else -- say, more of the Kotov/Dvoretsky/Stoyko type activity?

I think it is a combination of memory and active thinking, so both are likely necessary. And note MDLM used the first Circle or two to work on such things. The Circles don't claim to be a balanced approach to chess. They exclusively focus on tactics, to the exclusion of everything else. This is insane, and the Circles are not for everyone (I discuss who might consider doing the Circles, and how to avoid burnout, in this post).

Training beyond 4 or 5 hours in a day is counterproductive - by then the bucket is full until the training is consolidated during sleep. This argues against some of the later circles.

Is it counterproductive to practice a language more than 4 or 5 hours a day? It could likely have an "immersion" effect. Not sure what your claim is based upon. Also, if it is a concern it is easy to change the Circles so at most you only do a couple of hundred in a day.

Also, "attention" is a critical part of skill acquisition. If in the late cycles you take the approach of looking at a problem, and deciding "Yeah, I recognize the pattern...NEXT!" then I fear you may be rewiring your brain to be just as superficial. Properly done, each training position should be treated with the same level of rigor as a real game position: rough material count, tactical scan, candidate move selection, calculate, evaluate, lather, rinse, repeat, and blunder check. This point also argues against the later circles.

The goal is for a certain skill to be done automatically, without thinking. If you have ever trained at swimming or tennis you know the goal is not to eventually attend to your stroke, but to have such a good habit that you do it without thinking. At first you have to attend to how you are doing it, but eventually you do not. Indeed, once you are good enough when you attend to your stroke you do it worse.

There are certain chess skills that we should display without effort, things you should find even in the fourth game of the day in your tournament, when your mind is burnt and you can't calculate to save your life. Things you should see if you are giving a simul and have no time to think.

That previous paragraph is key, incidentally.

One such skill is you want to see basic tactics, you want them to pop out at you as vividly as your recognition machinery when you see a good friend's face in a crowd (of course in both cases you must double-check to make sure it is really him, or that the tactic is really playable).

There has to be "resistance." If you really can do Reinfeld's 1001 in a single day, perhaps it was well past time to pick up a more difficult exercise set. Now, one might contend that the speed requirement provides the resistance in this situation; well, if you go through a full thought process for each position, then perhaps one has a point. But even so, the speed requirement is artificial in this context: You develop proficiency by doing deliberate, conscientious, high-quality work, not by trying to do it fast. Speed then comes automatically.

See tennis/swimming analogy above. At first, attend, and once you are expert, you don't need to think. You just do.

Thus I would modify the approach as follows: Do a fixed number of exercises a day (or a fixed amount of time); repeat the set of exercises if they continue to be challenging, but move on otherwise. And forgoshsakes, study some strategy and endgames!

This seems perfectly reasonable. Ultimately it is an empirical question whether your technique would be better than the Circles. The chess improvement blogosphere needs a benefactor so we can do some experiments! (The most conclusive criticism, in my opinion, comes from Nunn, as I summarized here).

Saturday, February 07, 2009

Women are as good as men at chess

New paper out today suggests the top men are better than the top women because the population of male players is so much larger than the population of female players.

Bilalić, Smallbone, McLeod, Gobet (2009). Why are (the best) women so good at chess? Participation rates and gender differences in intellectual domains. Proceedings of the Royal Society of London (B), 276: 1161-1165.

A popular explanation for the small number of women at the top level of intellectually demanding activities from chess to science appeals to biological differences in the intellectual abilities of men and women. An alternative explanation is that the extreme values in a large sample are likely to be greater than those in a small one. Although the performance of the 100 best German male chess players is better than that of the 100 best German women, we show that 96 per cent of the observed difference would be expected given the much greater number of men who play chess. There is little left for biological or cultural explanations to account for. In science, where there are many more male than female participants, this statistical sampling explanation, rather than differences in intellectual ability, may also be the main reason why women are under-represented at the top end.

Thursday, February 05, 2009

Learning to Add, and some Rules of Conduct

Note: The Durham Underground Chess Club meets at 7PM, not 7:30PM on Tuesdays and Thursdays at Francesca's on 9th Street. Hope to see you there everyone! One reader has shown up and kicked my ass pretty much every game the past few weeks.

I had a second meeting with Coach B a couple of Saturdays ago. Just like the first, it was very helpful. Going over my games with a great player (that is also able to explain things--a key qualification) is much more instructive than reading over annotated games that GMs play.

One reason it is better to go over your own games is obvious: the games are yours: there is an emotional attachment so the lessons are more likely to stick. Second, GMs don't make the kind of obvious errors that my opponents and I make, so studying annotated GM games is like studying Calculus before learning how to multiply and divide. This is why I think GM-annotations of amateur games are so helpful.

Coach B also mentioned quite a few rules that he had learned from his coach. Here are some of the rules of the middlegame:
1. Never be surprised.
If your opponent plays a reasonable move you hadn't considered, you need to think things through more when you are selecting moves. It is a sign you are likely not calculating broadly enough, perhaps calculating too deeply.
2. Never play f3 (f6).
This weakens your King and produces pawn structure weaknesses.
3. Never sacrifice.
4. Never capture.
Let him take your piece, and develop your defending piece when you recapture.
5. Never move a pawn.
6. Always repeat once if given the chance, even if you aren't going for the draw.
This entices your opponent to blunder if he wants to avoid a draw, and if he is losing it gets his hopes up only to dash them to the rocks.
Obviously these rules have no exceptions and should be followed mindlessly.

One of the games we went over was very helpful in showing how important rook activity is in the endgame. In one endgame we were materially equal but my rook was stuck defending a couple of pawns near my second rank, while his rook was free to frolic and play at will. I was destroyed. Coach B reminded me that it is often worthwhile to sacrifice a pawn (or two) to have an active rook in the endgame. I've read similar things before, but it was helpful to have a case study from my own game.