Saturday, January 03, 2009

Chess for Zebras, Chapter 6 (Why chess is hard)

A summary of Chapter 6 of Rowson's Chess for Zebras. This is the first chapter in Part II which he calls 'A mental toolkit for the exponential jungle.'

Chess is hard. We all know this. The number of possible legal moves increases exponentially as we look ahead in the game tree. It is surprising that we feel we have any understanding of the game at all.

One way that people tame the exponential jungle is via heuristics that quickly trim down the game tree in our minds, focusing our attention on a small proportion of the total number of lines. One of the most useful concepts we use is that of material value: we don't bother exploring lines that would give away a Queen for free. There are also strategic ideas such as piece activity that filter out certain moves as not deserving consideration.

Such concepts we use to trim Kotov's tree can be useful, but in this chapter Rowson focuses on how they can be harmful.

Thinking in words versus thinking in chess
While it is often useful when a chess book explains a position in words, in practice much of our thinking during a game is not formatted as words, but as images. Words can be helpful in sizing up a position, serving to simplify the immense complexity in front of us, but they can also distract us from visualizing the concrete position, and blind us to the complexities that are actually crucial to see. Rowson says that GMs don't usually think in words nearly as much as amateurs.

Why are amateurs so attached to their word and rule-based thoughts? It is partly because amateurs have learned many fewer patterns than the great players, so the complexity of the position is overwhelming. To tame the complexity they use the only concepts they have at their disposal.

GMs, on the other hand, have learned many patterns and this eases the cognitive load on the visualization process. Their imagery is more abstract than the amateurs: they don't visualize individual pieces and where they are, but configurations of pieces and how these configurations relate to one another (e.g., most of us probably have such a chunk for the configuration of a castled King: Rf1, Kg1, f2, g2, h2). They don't imagine the pieces as colored, as having a certain shape, there is nothing inessential in their imagination. Recall my discussion of Binet's study of imagery in chess masters: I wrote that specifically to supplement this summary.

Rowson urges amateurs to think less in terms of words, more in terms of images. This will be a daunting task. How can you work on this skill? With a variant of Rowsonalysis of course! Rowson says:
To reduce the role of words in your thinking during play, you need to practice thinking about positions with as few prejudices as possible and observe your thoughts closely to watch for the pseudo-explanatory verbal 'solutions.' This can be quite a scary experience. Glimpsing 'the abysmal depths of chess' is highly worthwhile when done in moderation. If you can manage it, and if it doesn't put you off chess completely, it should help you to start building resistance against one of the main habits of mind that prevents you from improving.
This experience of staring into the abysmal depths of chess, free of our beloved rules and narratives, can be quite disconcerting, but it can also enhance your creativity and candidate move generation process ('What if I take his Bishop with my Queen and let him recapture?' is a thought you will be unlikely to have when thinking abstractly of 'material', but it could be just the right move in the position).

Rowson's take on Aagaard versus Watson
Recently Watson and Aagaard fought intensely about the importance, or lack thereof, of rules in chess. Watson says that great players play in a rule-independent fashion, while Aagaard has fought this idea vehemently. Rowson, to adjudicate the dispute, says that there are two senses of 'independent' that need to be differentiated. First, you can play independently of the rules if you are simply ignorant of the rules. A case in point is the beginner who doesn't know that Knights are poorly placed when on the rim. The second sense of independence is what Rowson calls rule-transcendence. Those that have transcended the rules have studied them well and fully absorbed them. However, they are good enough that they never accept a rule mindlessly and always consider the concrete position on the board.

Planning in pencil, playing in pen
Given that words and narratives are sometimes counterproductive during a chess game, what about planning? Plans typically involve concepts such as material, pawn structure, piece activity, etc.. Does Rowson think such heuristics we use to trim Kotov's tree are useless, that they should be totally abandoned? Should we live in the abyss of concrete chess, sitting there imagining variation after variation? No. Without plans we would get lost in the abyss. The point of the previous section is that we need to be careful of getting too locked into narratives that can blind us to important variations we do need to visualize.

So Rowson thinks planning is fine, and necessary, for good chess. He stresses that plans (which often use narratives) should be flexible and adapt to the particulars of the position. You should be ready to exchange the Knight you have on the great outpost if it will yield even larger advantages. An exchange that was a bad idea on a previous move might now be perfect given the change in the position. Plans are just ideas, so don't be afraid to change them or reject them outright. Moves, on the other hand, are permanent.

There is a Greek God Proteus who can change his shape at will. Rowson says that good chess players are protean, or flexible, in their thinking. Many chess authors say you should play 'consistently', or stick with your plans. Rowson tells us that consistency is the hobgoblin of little minds, that we need to constantly adapt our battle plans to the circumstances on the battlefield.


Blogger likesforests said...

"First, you can play independently of the rules if you are simply ignorant of the rules. A case in point is the beginner who doesn't know that Knights are poorly placed when on the rim. The second sense of independence is what Rowson calls rule-transcendence. Those that have transcended the rules have studied them well and fully absorbed them."

1900-2000 level players typically only know some of the rules well, so these players who have fully absorbed them are likely 2200+.

1/06/2009 08:29:00 AM  
Blogger BlunderProne said...

On Narrative versus concrete Chess:

Like we’ve discussed at Tempo’s before and other places, Learning chess is like learning a language. The Narrative is the crutch we rely on to “translate” concrete chess. True, knowing enough patterns and chunking is key to reducing the need to “translate” less. Recognition plays a key role in this. But in order to get there, we have to have our memory markers built up and time and again, a narrative approach is best.

Rules versus no Rules:

I think most of us can comprehend that in chess, for every rule there are exceptions. The trick is to not to turn narrative translations a blind dogma. Tarrasch, back at the turn of the last century made such rules and reinforced it with success to the point that a whole generation of players were following that style of play. It took a new movement from Tartakower, Alekhine, Nimzovitch and Reti ( others too) to offset this concept. Dr. Lasker, trained in the classical period of Steinitz and Tarrasch adapted to the new ideas because he saw the importance of the EXCEPTIONS to the rules. I will point out that the Hypermodern players mentioned above, strong as they were, never became world champions. Why is that? Perhaps they overcompensated too much to the point they were inflexible in style.

Which brings me to the last point I would like to make:

Flexible Planning:

All the great players became great because they learned to adapt to different styles. Kasparov was an aggressive, attacking player by nature but it was only when he adapted to Karpov’s more positional style did he gain some ground. Lasker did the same thing and so did Fisher. Flexibility in playing style starts in flexibility in individual game strategies. For instance, you may start out with an idea to have a king side attack as white in a middle game because you placed a knight on e5. But as black equalizes you realize that you may have to shift to a control of the center file plan. As Black defends the threat of a kingside attack, you see that the nuisance of the knight on e5 will eventually cause black to advance the f-pawn weakening the pawn on e6. So you build up a battery on the e-file.

I like this stuff! Makes for nice lunch time thought.


1/06/2009 12:21:00 PM  
Blogger Blue Devil Knight said...

LF: there is clearly a big gap between first learning some rules and getting to rule independence.

Upon reading this book I have screwed up more than a few games by finding pseudo-exceptions to the rules. My new rule is to follow the rules unless I can see a clear tactical refutation. It has been serving me well. I figure that is the best way to learn the exceptions: have them shown to me in real games.

The rules are there because they, on average, lead to better positions, so I've actually gotten better by following them. Not completely mindlessly, but I realized I certainly can't play as if they don't exist, and this chapter showed me that this wasn't Rowson's intent at all.

1/06/2009 12:46:00 PM  
Blogger Blue Devil Knight said...

The move from rule igorance to rule transcendence is not a step function, but a slow-rising sigmoid function . :)

1/06/2009 12:48:00 PM  
Blogger Blue Devil Knight said...

BP: a very good supplement to Rowson's book. I think he tends to focus almost too much on the negative impact of narratives and such. When I finish these summaries, I'll be writing an actual review of what I've read so far.

Overall I now can see why people like his book so much. It's great stuff.

In the next chapter he gives a positive story about the evaluation factors he uses (e.g., space or whatever). It will take me a while to write a summary, as it is now low priority, and it will likely be my last one (this stuff is good, but for reasons I mentioned above it isn't the best use of my time: I think if I were at 1800 or 2000 or so I might be ready). I still need to absorb the rules more before I can 'transcend' them. :)

1/06/2009 01:27:00 PM  
Blogger BlunderProne said...

I read the Zebra book 2 years ago. My history lessons are making more sense, filling gaps, teaching me flexibility, and learnign style. I will come back to Rowson once i am in striking distance of Expert.

1/06/2009 01:52:00 PM  
Blogger Temposchlucker said...

The stage of a narrative will not last very long in our mind. We simply can't help to transfer narratives into pattern recognition. That's how we are build.

The problem is: how do we formulate narratives of sufficient quality? That is not as easy as it sounds. The point is that both we ourselves and the ones from who we learn are biased. You have to work hard to get any objective information.

There is the rule and there is the idea behind the rule. You can sin against the rule while you don't sin against the idea behind that rule. Every idea has a scale of subtlety from very gross to very fine.

Further there is a set of rules needed that describe the hierarchy of rules. The hierarchy tells us when a rule can be 'violated' for the greater good.

1/06/2009 02:43:00 PM  
Blogger Rolling Pawns said...

Thinking in images rather than in words sounds interesting, I should try it. I think it makes the whole process more consistent, as finally you anyway have to go to images to practically calculate the move.
Thanks for adding me, I added link to you too.

1/06/2009 02:47:00 PM  
Blogger Wahrheit said...

Two points: First, Blunderprone, I love your series of posts on the great tournaments of yesteryear, you've been looking at these guys closely, but I think that Steinitz and, especially, my hero Dr. Tarrasch get a bit of a bad rap for being too rule-bound. As Yermolinsky and others have pointed out, it was in their writings for amateurs that they tried to "tame the exponential jungle" with a number of principles and rules. In their tournament games, they were big-time calculators of variations, as are all very strong players.

It is interesting to note that the hypermodern Nimzovich called his big book "My System," and that it seems to be full of...principles and rules. But in his games Nimzo was a creative unorthodox concrete tactician.

Second, I think that the Watson v. Aagard contretemps was quite interesting, but perhaps the problem is with the word "rules" and its different meanings. For example, a computer program is nothing but a list of rules--do this, if this then do that...but in chess the only time this might be said to apply is when there is only one legal move--other than that the "rules" are really just guidelines, e.g put your rooks on open files. This guideline represents a fantastic advance for the beginner, who otherwise will need hundreds of games to learn it by trial and error.

Yet, the guideline should read, in full, "put your rooks on open files if it is a GOOD MOVE." It is really helpful to look at the rook move, but one must then calculate to see if it's actually GOOD. GMs do the same thing at a higher level, based on their more extensive pattern recognition. They have a repertoire of ideas (or patterns) like, "If the pawn goes to h7 it threatens to queen and ties down his king and rook, and my rook invades the queenside." But they still have to make sure the move is good.

I see Rowson's overall vision for improving adult amateurs as getting us more flexible and yet more concrete in our thinking; more flexible in not focusing too much on a few rules that we've always relied on, and more concrete calculation of variations. But with the understanding, as in this post, that heuristics are indispensable in human play.

1/06/2009 04:20:00 PM  
Blogger chesstiger said...

In regards to the images i think it´s what Silman does with his imbalances thinking method.

First picture the position you want and then find ways to come to your dreamposition. Do the same for your opponent. If your plan is quicker then go for it.

Well that is in short Silman's thinking technique described to more detail in reassess your chess.

I guess i already am thinking in images, i only need to get those images more clearer to improve to the next level.

1/06/2009 04:55:00 PM  
Blogger Blue Devil Knight said...

Tempo: an interesting idea, which I know you have explored a lot (as have I with my 'explanation' (i.e., narrative) based approach to the Circles, which was clearly very helpful). I will talk about this more in my review which will probably be a couple of months :) I haven't sorted it all out yet, and probably never will, but I plan to be clear about what I'm confused about.

Rolling Pawns: Thanks for the link! I really like your blog.

I've been having fun just playing slow games focusing on how often I use words versus visualization. Typically I am visualizing, but the same couple of moves over and over, back and forth, as Kotov talks about (I gave the Kotov quote here, it is quite funny and makes me feel embarassed for how often it describes my thinking). Soltis thinks this 'back and forth' technique isn't always bad, for reasons I discuss here. For me, it is usually bad :)

Wahrheit: Excellent points, and I think you have captured Rowson's point very well.

CT: I can't wait for his new edition to come out, completely revised, as I mentioned here with a nice email from Silman.

1/06/2009 10:01:00 PM  

Post a Comment

<< Home