Why there cannot be a 'scientific' theory of chess
This blog is partly in response to recent blogs about Steinitz 'scientific theory'of chess and partly a response to responses to that blog.
http://blog.chess.com/kurtgodden/the-first-scientific-theory-of-chess
http://blog.chess.com/kurtgodden/steinitzrsquo-theory-of-perfect-play
Let me begin by asking some questions that I hope to answer in the blog:
1) Could there possibly be such a thing as a scientific theory of chess? I will try to show how there can be no such thing.
2) Can there be such a thing as a theory that helps a human being play chess better? Yes, there can be and are, provided we stick to a rather loose conception of what a theory is and focus on the idea that it is human thought and action as applied to the game of chess that is the focus, and not chess itself. Books on general guidelines and principles of strategy, for example, could constitute a theory in this loose sense. Opening books could be considered part of a wider theory of chess insofar as they are further articulations of a general theory. 'Theory' as used in common chess parlance seems to refer to what is currently considered the best move or at least a playable and studied move with some historical precedent (ie, it was played before, and seems to lead towards a draw, or slight advantage for white, etc). Similarly, there could be a theory that helps a computer play better, ie, the kinds of things you'd want to program into your chess engine to make it better. There may well be overlap between a human theory of playing chess and a chess computer program theory for writing a better program.
3) What is chess? On the one hand, we already know what chess is, just as we already know what justice is. On the other hand, we can get a greater clarity on what chess is just as we can get a greater clarity on what justice is by asking and answering more specific questions, just as Plato did in The Republic. I shall, in my own pathetic way, try to provide clarity on what chess is by comparing and contrasting chess to non-chess games, and distinguishing from the different aspects, elements and kinds of chess games, along with some comments on science and theory more generally. By gaining a greater clarity on what chess is, it will be shown and further supported why I answer questions 1 and 2 the way that I do.
First off, if Steinitz claims to have a 'scientific theory' of chess, I suspect he is using the term 'scientific' in the sense of the German word "Wissenschaft". A Wissenschaft is simply a general domain of knowledge about a subject. So you can have a 'science' of physics, a 'science' of art criticism, and a 'science' of chess. The term 'science' in this German sense is not at all meant to distinguish itself in the way that contemporary Anglo-American academics tend to categorize natural science (physics, chemistry, biology, etc, but not economics, psychology, sociology, etc.). "Scientific" here does not connote anything high and lofty and has nothing to do with the 'scientific method' (if there is a such a thing--read Feyerabend's interesting though controversial, "Against Method").
That said, perhaps there can be a 'scientific' theory of chess in the high and lofty sense that would be appealing to somebody who, say, wanted the 3 laws of chess just as Newton gave us three laws of the physical world. I will now argue why there cannot possibly be a natural science theory of chess in this more rigorous Anglo-American sense of 'science' as natural science.
Chess cannot possibly be the object of a natural science theory because chess is not part of the physical world. Chess pieces do not populate the physical world as chess pieces, they populate our ordinary, everyday world as chess pieces. Chess pieces and boards are superfluous to what chess is, just as our fingers are superfluous to numbers, though we can use our fingers to help us count and understand numbers. We can also just pick our noses. Fingers do lots of things, there is nothing essentially mathematical about fingers. Since nothing that is not part of the physical world can be the possible object of a natural science theory, chess cannot be the possible object of a scientific theory. You could, however, study the human brain when people are playing chess, but that will not be essentially a theory on chess, but merely on the brain (just as you could study what happens in the brain when we do math, that does not reveal anything essential about mathematics).
Natural science is in the business of giving us causal explanations of physical interactions and events in the physical world. Chess is not a part of the physical world. What is not part of the physical world cannot be the object of a causal explanation in a natural science sense of an explanation, and, therefore, cannot be the possible object of a natural scientific theory.
Chess moves, chess games, past and present and future, are not 'data' that can be manipulated to give a scientific theory that predicts anything essential about the game of chess. Data on the past human performance of anything can, however, be helpful in predicting, at times, future human performance of just about anything. But that cannot hardly reveal anything essential about what chess is. Let me further clarify the futility of and misunderstandings of trying to get a theory of chess by way of analogy.
Imagine you wanted a 'theory' of tic-tac-toe. You could analyze thousands of games of tic-tac-toe played by 5 year olds. Sometimes X wins, sometimes O, sometimes it's a draw. You can have a sophisticated rating system and tic-tac-toe books. You can stroke the egos of the very best 5-year olds and anoint them "Grandmaster" and even crown one of them World Champion of Tic-tac-toe. But as we know, tic-tac-toe, if properly played, leads to a draw. And any 'theory' on tic-tac-toe would simply help a (5-year old) play well. Chess, quite formally, is nothing but a sophisticated kind of tic-tac-toe. The main difference is that, at the moment, we do not know if what seems to be a first move advantage for white is sufficient to force a win, or if black can manage to draw. As another poster pointed out, there can only be three outcomes in the analysis, of any (legal) chess position, including the starting position. Either one side can force a win, or there must be a forced draw, assuming perfect play. At the moment, we do not know what the proper outcome of that analysis is, just as 5 year olds, we can imagine, would not know the proper outcome of the starting position in tic-tac-toe.
Chess, like tic-tac-toe, is purely formal. It is constituted by the rules that we made up. Change the rules (ie, get rid of en passant) and you have a new game. Just as you can change or eliminate any of Euclid's axioms and create a whole new 'world' of non-Euclidean geometry. Fischer chess (chess 960) does just that. It's not wholly different from chess, the way, say, rugby is different from chess, but it's a different game nonetheless. A 'theory' of chess that attempts to define what chess is could just as well be a statement of the rules that define it. A chess rule is not a causal law, it does not 'predict' anything, just as geometry doesn't predict that three lines joined together will magically create a triangle. That's just what a triangle is. Prediction presupposes past and future, that is, time, and chess is outside of time. Of course, you can lose a game on time, but that is the activity of playing chess that is temporal and your losing on time is a feature of those games of chess that are timed. The logical possibilities of chess games, in contrast, are, quite literally, out of this world.
Even though chess is purely formal, the activity of playing chess involves psychology, the occasional trash-talk of blitz, pre-move strategy in 1 0 games, and time-management. The activity of playing chess can have a 'theory' in a loose sense, and there are many books out there addressing various chess situations and issues. And again, we don't need to make any comparison of chess to physics or natural science or 'scientific theory' to understand and appreciate the value of the books already out there on these various topics (perhaps I shall write a 'theory of chess trash talk' and that would be as much a 'scientific theory' on chess as anything Steinitz wrote). The human activity of playing chess is quite rich because people are extraordinarily complex. But chess is fairly simple.
Tic-tac-toe is fun for 5 year olds because they just don't know to play it well enough yet. Similarly, chess is fun for us because we just aren't good enough to figure it out yet. That is not the conscious reason why we like chess, but it is a presupposition of our having fun playing chess that we, and our opponents, have not yet figured out perfect play.
I hope to have answered the questions above and, tangentially at least, some of the questions, issues, conflicts and confusions that arose in previous discussions by others in other blogs. It is indeed thought-provoking to consider what a theory is, what science is, how a theory relates to science, and whether there could be a scientific theory of chess. But chess cannot possibly be the object of a scientific theory in the robust sense of 'scientific', and there are already lots of 'theories' of chess (non-scientific though they may be) that do a good enough job at illuminating different aspects of the game.
The more interesting question is why anybody would even want a scientific theory of chess in the natural science sense of 'scientific'? The hankering for a scientific explanation for everything is itself a symptom of our misunderstanding our place in the world, a misunderstanding of the many dimensions of the world we live in (ie, the economic dimension, the psychological, the historical and cultural, the ethical, etc.), and a misunderstanding of what it means for us to have any knowledge at all about the world we live in. Natural science theory is like a hammer, and it tries to turn the world, in all its richness, into a dull nail. That a scientific theory cannot pound chess reveals more about the limits of scientific theory than it does about chess.
That chess fascinates us, and leads us to write books, hold tournaments, and anoint masters and grandmasters also tells us relatively little about chess but a lot about us. Namely, that we are overgrown 5 year olds with too much free time on our hands. But, like 5 year olds, we know how to have fun.
