Chess for Zebras, Chapters 2-4 (The logic of storytelling)

Here is a summary of Chapters 2-4 of Rowson's Chess for Zebras. There is so much overlap among these chapters that things will be more tidy if I lump it all together.

Recall that in Chapter 1, Rowson argued that chess is more a game of skill than of conceptual understanding. In this chapter, he discusses how our "knowledge" can actually harm our chess skills.


Storytelling
Many people incorrectly believe that chess primarily requires skills in deductive formal logic. While a computer can succeed via purely logical variation-crunching, humans are fairly slow and error-prone using this Kotov-style of reasoning.

For example, since humans can't exhaustively search the game tree, we have to use something other than pure logic to determine which moves to begin analyzing. Rowson says, "[L]ogic by itself cannot tell you which features of the position are most important, and therefore what you should be reasoning about."

So if we aren't logic engines, what are we? Commonly, we act as storytellers that construct narratives about positions and games, and these narrative arcs give individual positions meaning and context. One typical idea might be, "He created pawn-structure weaknesses around his King on move three, so from that move onward the winner looked for ways to exploit that weakness and go in for the kill." This one theme can be highlighted while explaining multiple moves and make us feel that we really understand the core idea of the game.

Unfortunately, people's narratives often harm their performance over the board, and chess is a game of skill, not of understanding. People's preconceptions about chess, and themselves as chess players, engender narratives with detrimental consequences . In an excellent summary, the Brixton Chess Club blog says, "We tell ourselves a story about [the position] and we consider the board in the light of the narrative rather than the objective truth."

So, should we try to clear ourselves of the narrative urge, stop reasoning in terms of stories and plans about the position? No. We are story-telling creatures and this is part of what makes chess so much fun. What is needed are better narratives, not zero narratives.

Narrative Pollutants
Before discussing what a good narrative would look like, let's examine a few examples from Rowson's taxonomy of narrative pollution.

Opening narratives
Standard narratives often accompany a pet opening line. If your opening has a typical set of moves, those moves start to feel "right," to the point where you overlook better moves. If you typically castle queenside in an opening, for instance, you might overlook cases in which that would be a mistake.

Rowson has a funny story about this. After beating a cocky player, in the postmortem Rowson suggested his opponent should have castled. Rowson says, "My opponent looked at me like I was a complete idiot and said: "No black does not castle in this line." [H]e had become rather too attached to the idea of not castling and had rigidly associated that idea with the opening, thus limiting his legitimate options considerably."

Chess identities
Our stories during a game are also strongly shaped by our picture of who we are as chess players. There are several common mythological characters with whom people self-identify. One, the 'sacrificial attacking player' will often overlook strong quiet moves, or take unnecessary risks that severely weaken their position.

There is also the 'noble apprentice', who doesn't deign to get emotionally involved in games of chess. He dispassionately looks at each game as a learning experience, especially when playing against higher-rated players. Rowson says, "Be careful with the idea of learning from your defeats. If we become comfortable with the idea of learning, it can undermine our will to compete." Sure, learn from your losses, but get in there and fight to win! Don't protect yourself from the agony of defeat by sacrificing your competitive spirit.

There is also the 'thwarted genius', who places such a high value on their intellectual understanding of the game that they don't have particularly good skill at actually playing. They are often good at pointing out errors in people's games after the fact, and will often try to "outplay" their opponent, hoping for an impressive win rather than simply a win. Rowson says, "Don't try to 'outplay' your opponents; just try to beat them!"

Overall, identifying too strongly with such mythological characters screens us off from seeing objective features of the position that are important. Rowson says, "The primary effect, I think, is that when the myth imposes itself on us, we try to impose ourselves on the position."

Good Narratives
The common denominator with counterproductive narratives is that they are constructed without enough sensitivity to the complexities of a position. It is all too easy to take in one or two facts about the position and integrate them into an overarching grand narrative. We end up with vague narratives such as "His Kingside is weak, I'll attack!"

In contrast, one of the main properties a good narrative has is that it is specific. Instead of "His Kingside is weak", how about "The f7 square is unprotected." Instead of "I'll attack" why not "Perhaps it is time to attack. I might move my rooks to the semi-open f-file to put more pressure on his King." A helpful narrative often mentions specific squares or pieces.

On a related note, good narratives typically involve short-term plans, not grand long-term plans. Rowson says, "[I]t is wise to be extra suspicious of any plan that takes more than three moves to implement. It's not that you can't make long-term plans, but just that your opponent will usually obstruct them in some way, so anything that takes more than two or three moves at a time to complete requires an extra-careful look to see how the opponent will intercept your idea."

In addition, a good narrative should be flexible and open to revision. While playing a game, the outcome of the story is still not set. We should not get attached to any particular story we are telling ourselves (especially be wary of those that carry over from several moves ago). We also need to remember we have an opponent, another author contributing to this story, and he is trying to sabotage your work.

Perhaps most important, the position has a story to tell, so we need to spend time reading the story of the board rather than writing it. Rowson advises, "[Y]ou need to leave space for your opponent's version of events, and also think of the story the position is trying to tell you, before letting your own story dominate your thoughts. If you manage to do this, your assessments should become a little more tentative, and this usually helps you to make better decisions."

One of the main ways to find holes in the plot, i.e., defects in the story we are telling, is to work through concrete variations that will result when you play it out. As for how to analyze such variations, Rowson discusses the study I talked about in depth previously on my blog. The main result is that high-level players look for worst-case scenarios when they considering a move. They pick at the move like a nagging stepmother, trying to come up with their opponent's best response. This tends to yield a fairly objective evaluation of the move. More novice players, on the other hand, tend to consider responses that are weak, that confirm their prejudice that a particular move is a good move.

The better a player is, the less confident he tends to be in his initial evaluation of a position, saying things like, "White looks better here, but we'd have to analyze it." This humility in the face of chess complexity tends to make them more thoughtful in their analysis of a position. Rowson says, "It is much wiser to approach chess with an "I don't know; let's see" mindset, than an "It's like this; I'll prove it" mindset."