The Bobby Fischer Defense, by Garry Kasparov
nybooks.com
nybooks.com
"Brady’s dismissal of this theory misses the point: 'What everyone seemed to overlook was that at the board Bobby feared no one.' Yes, once at the board he was fine! Where Fischer had his greatest crisis of confidence was always before getting to the board, before getting on the plane. Fischer’s perfectionism, his absolute belief that he could not fail, did not allow him to put that perfection at risk."
That applies to much more than just chess. In fact I found the whole article not just very well written but poignant on many levels.
(It doesn't entirely fit Fischer, who seems to have done most of his going-mad after retiring from his chess career rather than during it, but never mind.)
...Fischer’s modern interpretation of “victory through clarity” was a revelation. His fresh dynamism started a revolution; the period from 1972 to 1975, when Fischer was already in self-exile as a player, was more fruitful in chess evolution than the entire preceding decade.
We need some more of this objectivity in programming. The programming domain is so complex, it's easy for people to create dogmas and defend them with verbal salad. Some more "victory through clarity" would be beneficial.
I've believed for years that most of our great programming talent is completely unknown and completely unheard-of, working on antiquated database systems for some company somewhere.
Programming "culture", so much as it has one, could benefit a lot from arguing less and doing more.
Unless we share data about what we're doing, there's no way to grow this "culture." A way of arguing less and accomplishing more would be to share more objective data.
"Then on January 17, 2008, he died in Reykjavík after a long illness for which he had refused treatment. Even this was somehow typical of Fischer, who grew up playing chess against himself since he had no one else to play. He had fought to the end and proven himself to be his most dangerous opponent."
It seems like Fischer was playing against himself all his life. That being said, I don't think today's cold, database driven world of chess (e.g. see http://www.newyorker.com/reporting/2011/03/21/110321fa_fact_...) would have had much use for his individuality, had he stayed the course. He probably would have lost to the new breed of players.
Good for him, then, to stay a legend, albeit a disturbing one.
The general consensus among chess grandmasters is that the best algorithms out there are undeniably impossible for a human to defeat -- but yet, they play "ugly" chess. They're using a brute-force style of AI and can beat a human player down through sheer speed of calculation, and/or perfect recall of vast databases of former games. Someone like Kasparov would argue that a computer chess master is no more the intellectual superior of a human master than a calculator is the intellectual superior of a human mathematics PhD.
Finally, the article did confirm a suspicion held among some AI theorists and neuroscientists these days, which is that a combined human/AI "team" can defeat any lone human or lone AI. In the case of folks like Carlsen, the thinking is that this new generation will soar to new heights precisely because it has trained with and against computers, thus piecing together a best-of-both-worlds approach.
Except that there's a lot more to being good at math than just being good at calculation and memorization; not so in chess (at least not against computers).
Computers have proven that they've got what it takes to win the game. And, ultimately, that's all that really matters in chess as it is presently designed.
Of course, there are plenty of chess variants (not to mention games like Go) which are a lot harder for computers to play well. It is quite likely that that's where most serious chess players will migrate to over time.
In fact, for the last decades of his life, Bobby Fischer himself refused to play anything but Fischer Random Chess (also known as Chess 960), a variant which made it practically impossible to memorize a significant portion of opening theory, since in this variant there is not just a single opening position as in regular chess, but 960 of them.
I'm nowhere near good enough to pretend to understand the motivations of a chess master, but this does kind of make sense to me. Somehow, it kills the vibe to know that computers will always win.
Funny, humans still enjoy sports even though a human sprinter will never run as fast as a motorcycle. Maybe it's because we've always been physically less capable than animals, maybe it hurts more to be bested intellectually. Horses have always run faster than us, so we're primed to see the motorcycle as a tool rather than a challenge. In sports, maybe we're better at celebrating human achievement without comparing it to something else.
Like most people, I don't actually think humans really were bested by computers in chess. There is a huge difference, of course - the computers are winning by following a completely different "thought" process than a human. It's an astoundingly impressive accomplishment to build a program that can beat chess masters, but because the way a human plays is so different, I think that I understand what the calculator analogy is getting at. In some ways, the "victory" shows the severe limitations of computers - a human chess grandmaster, capable of holding and evaluating maybe a dozen positions, is capable of playing evenly with a computer that evaluates hundreds of millions of positions a second. And in Go, that dozen or so positions combined with intuition, strategy, and other mental processes we consider almost subjective is so effective that a billions of positions a second computer can't even come close to winning.
If go eventually falls because computers are now capable of evaluating trillions of moves per second, but the approach is still essentially number crunching, well that's an impressive engineering feat, but "human intelligence" won't be any more defeated than it was the first time, with chess.
Now, if a computer can beat a human by evaluating fewer moves, ok, maybe then we've been bested.
Kasparov seems to be an optimist, for what it's worth, and in several of his disquisitions on this topic, he argues that a computer program will never "solve" chess the way a program solved checkers a few years ago. There are too many numbers to crunch. Furthermore, there are too many weaknesses in AI's ability to form strategies or think creatively. Unfortunately, this line of reasoning assumes that creativity is necessary for success in chess. It might be in a human-vs-human game, but in absolute terms, evidently it's not.
"What Go is to philosophers and warriors, chess is to accountants and merchants"
If so, why is it not as popular as chess (outside Asia)? This could be examined at length but the fact that chess was used as one of the weapons of Cold War and its long history (Go has become popular in the West only in late 20th century) all should be considered. Yet, there might be some truth to Nicholai's (aka Trevanian, aka Whitaker) observation. The antagonistic nature of chess is much easier to grasp than the cold, crystalline-like beauty of Go's patterns.
As to your comment " I don't actually think humans really were bested by computers in chess", which is shared by others on this thread: The observations that computers play "ugly chess", the "thought" process is different, etc. stem from an romantic idealization of how the grandmasters are raised and train nowadays. I would very much suggest that you read the NYer article I linked (an another commentor summarized). The basic idea there is not that computers have learned to play an ugly version of brute force chess but humans (with the aid of their computerized tools) learned how to mechanically optimize their play, getting closer to computers.
Sure, but are the long term structural patterns in Chess easier to grasp? I'm at a disadvantage here because I'm honestly not good enough at either game to really know. I read that Bobby Fischer once snapped to a journalist who congratulated him on a well played chess match: "how would you know?" I don't know. My casual dabbling in chess and go gives me as much in common with Go or Chess masters as my BA in Math gives me with Euler. What I do isn't a weaker version of what they do, it isn't related at all.
Now, my 5 year old loves the antagonistic nature of chess, it's about taking the other guy's pieces. Kabam, got you! But c'mon, even I know that's not meaningful chess.
And these comparisons of Kasparov's genius to "accountants and merchants?" Kind or reminds me of that scene in "A Beautiful Mind" where Nash loses a go game and claims "the game is flawed." It's not flawed, you just don't get it.
I don't think most chess players will migrate to Go. It's the serious chess players which will probably migrate, and they'll probably migrate to a variant of chess rather than to Go.
Go is just too different for most players who are fascinated with chess to switch to (though they might appreciate both).
Many chess variants, on the other hand, tend to be similar in many ways to regular chess. So if you enjoy playing regular chess, you'll likely enjoy playing a variant just as much (maybe even more).
In fact, "regular chess" is itself actually a chess variant, as the rules of chess have changed over the centuries. What we call "chess" today differs in some ways from what was known as "chess" even just a couple of hundred years ago. And the further back in time you go, the more differences there are.
Chess is not a game whose rules are set in stone for all time. It changes and develops with the demands and interests of the ages. And it seems likely that the demands of our age will change it once again.
"I'm nowhere near good enough to pretend to understand the motivations of a chess master, but this does kind of make sense to me. Somehow, it kills the vibe to know that computers will always win."
It's not so much that humans somehow resent the dominance of the machines in chess, but that computers (and the study of chess in general) have radically transformed the game in a direction that many players are unhappy with.
For instance, it used to be that if a tournament game lasted too long, it would be adjourned (ie. the players would take a break for a while) until the game could be resumed. In the meantime they could think over the adjourned position and use what they thought up when the game resumed. Because of computers, that's no longer done.
The problem is that either player could now secretly use a computer to analyze the adjourned position for him, and from that point on the result of the game becomes more of a matter of who has the strongest chess computer rather than of which human is the strongest chess player.
Another problem that computers (and the study of chess) have brought about is that the memorization of previously analyzed moves has become a major part of the game (especially at the more competitive levels). This was true even in Fischer's day, before chess computers were useful, but is much more true today.
The consequence of this is that the person who's memorized the most chess moves can have a great advantage (though this is a bit of an oversimplification, as many other factors enter in to play as well). Many people are really not very fond of playing against what amounts to a database of chess moves rather than against another human being.
That's why chess variants like Fischer Random Chess are appealing: to a very large extent, they take away the usefulness of memorizing chess moves.
Of course, one of the advantages that computers have over humans is that they have a "perfect memory" (though not a limitless one). So when playing a chess variant like Fischer Random Chess, the playing field is leveled against computers as well.
Another problem that computers have brought to chess is the possibility of cheating using computers. Accusations of cheating with computers have plagued even the highest levels of chess, and certain types of chess ("postal chess" and internet chess in particular, where you can't see who you're really playing against) are very vulnerable to such cheating.
Playing a variant of chess that's harder for computers to win would solve this problem as well.
"Funny, humans still enjoy sports even though a human sprinter will never run as fast as a motorcycle."
Well, the thing is that a motorcycle doesn't actually run.
It'll be interesting to see what happens to sports like sprinting and boxing when humanoid robots get advanced enough to be able to actually compete in these sports.
I don't think they'll be quite as catastrophically affected as chess has been, since study with the robots isn't likely to give nearly as much of an advantage to players in those sports as studying with chess computers has given to chess players. But it'll still be interesting to see how they react once robots can actually beat them at their game.
http://www.nybooks.com/articles/archives/2010/feb/11/the-che...
1. http://www.rybkachess.com/ 2. http://www.cruxis.com/chess/houdini.htm
This is the same Karpov that a few matches earlier ceded that Spassky would win that current candidates cycle, and then Karpov goes on to beat Spassky quite comfortably in the semi-final match.
I think this speaks for Karpov's public-facing modesty and pragmatism, rather than an objective examination of playing strengths.
But still, a Fischer-Karpov match could have been absolutely epic, perhaps potentially surpassing the previous Spassky-Fischer match.
Actually Kasparov's grasp of reality is somewhat questionable. For instance read http://www.revisedhistory.org/view-garry-kasparov.htm where he demonstrates an astonishing lack of understanding of basic history. (For the first major mistake, see http://en.wikipedia.org/wiki/British_Agricultural_Revolution for why his naive projections of population growth are clearly wrong.)
In this particular case a lack of knowledge of British history is a lack of knowledge of world history. The British Agricultural Revolution resulted in major increases in food supply, that allowed populations to grow rapidly. What Britain discovered got exported to the rest of Europe, and European colonies. The result was a greatly expanded carrying capacity for the human population, which enabled a world-wide population explosion.
Now granted, most people don't know about this. But if you're curious about the history of population growth in Europe, it really is easy to find out what historians have to say about it. And when you do so you should run across this fairly quickly.
It's not quite the same thing as having a tenuous grasp on reality. It's more like listening to Einstein (or Chomsky) on politics -- very good in one field, rather over-confident in another.
Chomsky is very careful to heavily reference and back up his political opinions, so I'm a little confused as to how he is over-confident in the field? I assume you simply disagree with him?
Just like a good (in the sense of skill) salesman can sell you a product you don't actually want by producing all the evidence you need...
In the movie, his life was pretty depressing to see. Much like other child prodigies who have a great beginning, years down the road you're wondering what ever happened to them.
His awkwardness and ignorance had to have a great affect. I imagine a lot of it came from not having a father figure (or guidance in general) and becoming entirely consumed by chess at such a young age. He really closed himself off in his own world.