The Wall Street Journal's Five Best Chess Books
Not really the WSJ's choices per se, but Gabriel Schonfield's, the guest columnist for this past Saturday's "Five Best" column. His choices are pretty trite, but that's probably appropriate for his audience:
1. Fischer's My 60 Memorable Games
2. Kasparov's Garry Kasparov on My Great Predecessors [Great title: how else would we know that it was Kasparov writing about his predecessors?]
3. Tal's Tal-Botvinnik, 1960
4. Nimzowitsch's My System
5. Lasker, Lasker's Manual of Chess [Evidently one of Kasparov's great predecessors in multiple senses. Ah, marketing.]
There's some nonsense in his review too, like his claim that Tal's victory was "a demonstration that chess can be scientific only in the way that Soviet socialism was scientific, which is to say not at all." No argument about Soviet socialism, but one wonders if Schonfield has noticed computer chess. Also, how about the 9(!!) point swing in the 1961 rematch (Botvinnik lost 12.5-8.5 in 1960 and won 13-8 in 1961), or Tal's own claims in 1979 that, in light of his increased understanding of the game, he'd "tear to pieces" his earlier self?
On balance though, it's a nice little article for the non-chess playing public, and it's useful for chess players unfamiliar with those books, too. I don't think Lasker's Manual is especially invaluable, but I'd wholeheartedly recommend the other books to all my readers.
1. Fischer's My 60 Memorable Games
2. Kasparov's Garry Kasparov on My Great Predecessors [Great title: how else would we know that it was Kasparov writing about his predecessors?]
3. Tal's Tal-Botvinnik, 1960
4. Nimzowitsch's My System
5. Lasker, Lasker's Manual of Chess [Evidently one of Kasparov's great predecessors in multiple senses. Ah, marketing.]
There's some nonsense in his review too, like his claim that Tal's victory was "a demonstration that chess can be scientific only in the way that Soviet socialism was scientific, which is to say not at all." No argument about Soviet socialism, but one wonders if Schonfield has noticed computer chess. Also, how about the 9(!!) point swing in the 1961 rematch (Botvinnik lost 12.5-8.5 in 1960 and won 13-8 in 1961), or Tal's own claims in 1979 that, in light of his increased understanding of the game, he'd "tear to pieces" his earlier self?
On balance though, it's a nice little article for the non-chess playing public, and it's useful for chess players unfamiliar with those books, too. I don't think Lasker's Manual is especially invaluable, but I'd wholeheartedly recommend the other books to all my readers.
Posted by Dennis Monokroussos on
Monday March 24, 2008 at 11:08am
Since Gabe is a prominent neocon intellectual (in his day-job he's a senior editor at Commentary), his desire to upend Soviet socialism's pretensions to being "scientific" is likely ideologically motivated, at least in part. And of course you're right that later events such as Botvinnik's big margin of victory in their return match cast doubt on the very broad conclusion that Schonfeld drew from Tal's victory in the first match.
Nevertheless, you're wrong to call his point "nonsense." The first argument you raise, about computers, is off-base by 180 degrees. The ascendancy of computers represents the strongest possible argument against "scientific" chess (of the sort that Botvinnik championed, anyway); computers are far closer in spirit to Tal's purely emprical approach. Science is "empirical" only in the sense of using data to test hypotheses, which are supposed to grow out of theory, principles, systems. Being at root an effort to discern rules and principles (not just in chess, but in everything), science is the opposite of pure trial-and-error, which is how computers come up with the best chess moves.
I've debated this question with various anonymous posters on Daily Dirt ... but never with a real-live philosopher. So I'm curious how you'd respond to the issue of whether computers follow anything like a scientific approach in chess, based on a well-developed conception of the scientific method.
As for the list, I don't agree that the list is trite, although some of the comments about them seem to be. I've got more of a problem with the fact that one of his books is actually a five volume set of books, each individual volume of which probably contains more data than any of the other books. (I recall that the Petrosian + Spassky book was smallest, and Fischer's book is probably the 'largest' of the others.)
It's not clear to me that there really is any such thing as the scientific method, at least not in the textbook sense of observation-hypothesis-experiment-result-conclusion. All observation is already theory-laden, results are often ambiguous, the conclusion is sometimes to (rightly) reject the experiment and its results, etc.
The more important question is how Schonfeld is using "science" - in some way, it's supposed to be something that Botvinnik exemplified and Tal didn't. So what's that supposed to be? "Unfettered imagination"? There's no way Tal could become world champion without putting some restraints on his ideas, and there's no way Botvinnik could have become world champion and one of the world's best players for at least 35 years without a tremendous capacity for fantasy over the board.
So really, I don't think Botvinnik is a "scientist" in Schonfeld's sense or that Tal isn't. But let's try something else in the ballpark. Perhaps the "scientific" approach, i.e., Botvinnik's approach, means playing according to objective evaluation based on the demands of the position and not speculating. On this interpretation, the computer is a perfect exemplar of the "scientific" approach. Computers don't have fantasy: they just crank and crank and spit out the best move their algorithm produces. This is completely opposite Tal's "2 + 2 = 5" approach, and what I had in mind in my post. It's what I took to be Schonfeld's idea when he wrote of Tal's imaginative approach.
As for the trial-and-error approach of computers being opposed to a "scientific" empiricism, remember that Schonfeld is defining science in terms of what Botvinnik did, not what computers do. So unless you think that Botvinnik came up with his moves through the blind trial-and-error application of a rigid algorithm, the argument is a red herring. (But only in this context; in a normal context, it's very much worth considering.)
Mind you, discussion of "what is scientific" can devolve into arguments of terminology. My own feeling about how "scientific" the various top computer engines are will scale with how closely they agree with each other---in runs now underway here with Toga II and in Reading, UK with Shredder 10, for starters. My own working hypothesis is that agreement scales upward with Elo rating---i.e. as skill progresses, inferior "fantasy" gets squeezed out.
In particular, I conjecture (and frankly hope) that programs run to 15-ply in 10-line mode, which I place at Elo 2900--3000, score above 70% on my agreement scale, whereas players and programs in the 2700--2800 range agree under 60%. Maybe I'll need Elo 3100 or 3200 for the 70%. But if programs with different "personalities" disagree 40% and still reach 3200, then I'll conclude that "unscientific" factors of "style" still have as wide a scope as between (young) Tal and Karpov. My partner in Reading is also testing games with both players 2400 +/- 25, and we (falsifiably!) hypothesize that they will agree with the 3000 computers no more than 50%.
Had Tal understood the importance of being physically fit for Chess which was by the way obvious to Botvinnik and Fischer, his nonetheless impressive career would have overshadowed Botvinnik.
It is also interesting to note Capablanca, Fischer, Botvinnik, Karpov the leading practictioners of the scientific style defined by a clear logic and perfection of play have all predicted the demise of Chess. We now understand their prediction is taking the form of a machine that has preempted their style of perfection.
The romantics are still alive and well as we can see with Topalov, Shirov, Morozevich that have picked up the flag from Alekhine, Tal and Kasparov. The romantics never held such fatalistic views because they don't play like a machine. There is still room for creativity as we see Topalov play moves such Nxf7 that can still blind the Silicon beast with his shortened horizon. The Silicon Beast for now still fails to find those moves unless you take it to the edge of the cliff and get it to peer into the abyss.