Recently I picked up a book called How Life Imitates Chess by Garry Kasparov. I’ve found this book to be rather interesting, contrary to what one might expect from a ‘chess book’. Though I feared the book would be too domain-specific to chess-related things, I’ve found it to be rather general in nature.
How Life Imitates Chess is mainly about strategy, rationality and decision-making. Obviously chess plays a large part in it, but rather than being a tedious recycling of Kasparov's past life of chess, I've begun to appreciate the game. Before Life Imitates Chess, I thought of chess as a game where only computational capacity would really matter, making it seem rather trivial and not worth investing the time to learn it properly ("how could a game that is 'trivially solved' by a computer be of any interest").
Now, I see chess as a probabilistic game waged between the minds of competing the players, kind of like poker. Chess is complex. Perhaps perfect play always results in a draw, but no one plays perfectly. We play chess by reasoning with uncertain information about our opponents, and even of our own mind.
Though over 50% of chess games by grandmasters result in a draw, repeated games give us information about our opponents: information we can use to infer their strengths, styles, biases and other weaknesses. By exploiting this information about our opponents, taking into account our information about our respective attributes, we are truly playing mind against mind.
One aspect that makes How Life Imitates Chess so fascinating to me is Kasparov's vast amounts of experience on deliberate practice, practical chess play, on how creativity and one's talents and dispositions work, to some extent, and on how to make proper use of them.
Ultimately, chess has only one goal: to checkmate the opponent's king. Since we cannot compute the optimal moves from the opening, we need to find ways estimate our position's strength and to find practical approximations to 'optimal' play.
Kasparov introduces some of the broad patterns that have been discovered by chess players over time. In chess (and elsewhere), one's assets at hand can be perceived in terms of their values in Material, Time and Quality:
Material is the sum of the, well, material values of your pieces. Computers make generous use the material values to estimate the utilities of non-terminal states. As an example of how material values might look like, in chess the pieces are worth roughly the following: A pawn has value 1, knight and bishop 3, rook 5 and the queen 9.
In chess, besides the real-world limits on time use, there is (another) fundamental kind of time we have to be aware of: the board time. Time in chess is divided into moves and half-moves. For example, enhancing a position with a piece may net you some material advantage, but also has a cost in time. This cost may be enough to lose one's initiative to the opponent, which is one of the time-governed assets.
Quality is the way your pieces have been developed. A piece may have a certain, roughly constant material value, but when you take into account the piece's (lack of) mobility, how well it is positioned for, say, threatening the opponent's king, etc. The way I understand Quality is in terms of information. Suppose the pieces' value has a normal distribution with relatively low variance (or some "similar" distribution). To fully encode the value of position, stick to the normal material values, but, in case of uncommon or exceptional circumstances, use more bits to encode the fact that a piece's value has been shifted by a few standard deviations (say) from its default value.
Material, Time and Quality then, are the stuff of chess, and of life. One has to constantly balance, exchange and evaluate one's standing in those terms. Recently, I see economics playing out everywhere, and I wonder, how does economics govern these exchanges between, say Material and Time for Quality, or Time for Material, etc.? (the book does mention various such transactions being the equivalents of making long-term investments, etc. but I've yet to internalise that)
Kasparov emphasises the importance of recognising one's personal thinking style and fully exploiting it. Naively, I'd thought that one should be able to adopt any style whatsover, but those styles are rooted in various difficult to change factors. Kasparov says that two grandmaster-level players with very different playing styles are often
evenly matched. I'm not sure how much one could generalise this observation about the inconsequentiality to different thinking styles, but provisionally, at least, I will have the idea of styles as some elusive personal touches to the efforts of players to exert similar amounts of optimisation power to their positions.
Besides just thinking styles, Kasparov draws attention to 'talent' and other patterns in our lives. Talent, he says, comes from our predispositions for various skills, such as kinesthetic, various forms of memory, even self-discipline (which I don't often hear of as being included as a kind of talent).
Most talents require predispositions to multiple things. It is a priori unlikely to be 'naturally skilled' at any given endeavour, so it is important to expose oneself to as many different endeavours as possible. Ricardo's Law of Comparative Advantage comes to mind.
In the past I used to play games by proceeding with something like my first guesses at what might be a good thing to do. Poker, chess, RTS games, etc. were all quite like that. I would not think much about what I would do; I wasn't learning deliberatively; I was hardly ever improving. As Kasparov puts it, the desire to be quick and to keep moving forward overwhelms us, and prevents serious improvement.
Decision-making and search:
Kasparov makes the point that to make good decision, we need to explore the right amount of options to a reasonable depth. Usually this means a handful of options (usually 2 or 3 moves in chess) which need to be explored. It could be that some options lead to inevitable catastrophes, so always consider at least two options, and allocate enough time for each of them to be able to consider them in depth.
As with other things, I still at least half-held the naive view that I should have all the time in the world the carefully analyse 'all the data in the world' before I make my choice. Of course, this is totally unworkable, for reasons Kasparov mentions himself: there are time constraints, and, perhaps more importantly, over-focusing on the
information and its analysis easily leads to neglect of one's weaknesses in the actual decision-making processes. I am definitely guilty of this in my recent endeavours and otherwise.
Finally, my motivation in reading all this: I took interest in chess and poker in attempt to learn strategies and ways of thinking that would generalise to other things in life, as much as possible. Humans are, however, notorious at compartmentalising their lives, being extremely rational in certain domains while being lousy in others. I will have to think about ways to ensure that the skills would be readily transferable to other domains.
This is it for now. I will hopefully have improved somewhat on overall writing skill, but I think I'm still not able to give rise to a coherent, intentional structure to what I write. In the follow-up to this, I will try to explicitly look for the generalised real-life lessons that can be gained from How Life Imitates Chess, and possibly to inject how they cross over with my previous thoughts.
I may eventually even rewrite this post, provided I learn how I should begin to do so. For that, I will also need to learn to read better, and to store my notes in a more accessible form. As a result of my in-parts poor reading style, some of my references to the book may be inaccurate.