Friday, January 18, 2008

Bobby Fischer (Guest Post by Ken Regan)

When I watched the Fischer-Spassky match on TV back in 1972, there was a young chess prodigy, just a few years older than me, helping with the commentary of the games. That kid grew up to be Complexity theorist Kenneth Regan and I asked Ken to give some personal comments on Bobby Fischer, who passed away yesterday.

Bobby Fischer lived 64 years, one for each square on the chessboard. Unfortunately most of those squares were empty of playing the game he loved at the highest levels, but the brilliance and spirit of his games and ideas will ensure his board is remembered as more than half full.

What impressed me from Fischer's games was that clear logic and dynamism can both be harnessed. He produced scintillating attacks of the kind we associate with Tal and Kasparov, and positional masterpieces worthy of Capablanca and Karpov—including Game 6 of his 1972 match with Spassky. Kasparov is known for researching new ways of sacrificing pawns in the opening to increase the energy of one's position, while Fischer always maximized the potential of the position to hand. Almost uniquely with him there were no early draws while there was fight left. Other champions are known for how many years they went without losing a game, whereas Fischer won 19 games in a row, in the world championship stages. This ethic rubbed off on me even when I found myself paired against a fellow teen master I'd just shared a long bus ride with from Princeton to NYC. He sensibly proposed an immediate draw so we could rest, but I was there to play—and I lost!

I was attracted to the game just before the "Fischer Boom" years, and had nearly reached master level at age 12 when the Fischer-Spassky match began. I was on the nationwide PBS live broadcast of two of those games as an expert commentator assisting Shelby Lyman's TV coverage. My own rise was aided much more directly by the tournaments organized on a nationwide scale by William Goichberg, who is now President of the US Chess Federation. I never played Fischer—I met him only once in an elevator when he visited one of those tournaments. I was the age to feel the letdown most deeply when he did not play after 1972 and did not defend his title against Karpov in 1975, though what we've learned about his personal travails since then removes blame and much regret over this. Still, a piece of Brooklyn died with me yesterday.

Fischer will also be known for innovations of "Why didn't anyone else think of that?" caliber. The Fischer Chess Clock is now standard equipment. It regulates a player's time allotment in the manner of Social Security so that each move always has some thinking time. The Fischer Castling Rule enables the game to be started from different initial configurations while retaining its character. Fischer Random chess has both players start with the same random choice from 960 placements of pieces on the back row, and is gaining traction as more people agree with Bobby that computers and vast encyclopedias of opening analysis are causing the standard opening configuration to be "played out." I favor "non-random" placements with Black allowed to differ from White in my proposal Baseline chess with Fischer rules. Fischer also feared that computers would ruin the mystery of chess, but I can personally vouch that the game's incredible complexity remains. In response to the challenge compliment from Grandmaster Susan Polgar's premier chess blog, I undertook to tell whether (now ex-) World Champion Vladimir Kramnik missed a win at Move 50 on the slippery slope to losing his title last September. After four months and 300+ pages of analysis, aided by Deep Fritz 10 and two other chess programs running on faster hardware than DF10 used to beat Kramnik a year ago, having sifted over 20 trillion search nodes and tried out almost 100,000 moves, I'm about to throw up my hands and say I have no opinion more definite than "flip-a-coin" on whether White can win!

6 comments:

  1. fischer random is such a challenge! every time i play, i realize how formulaic and weak my opening game tactics actually are. leave it to bobby to come up with something so radically different. as you said, it's too bad we didn't get to see him more.

    btw, any one play online at freechess.org?

    -saket

    ReplyDelete
  2. It seems to me like the great chess player died a long time ago, leaving only a particularly odious human being. I can't say I'm sorry the latter is dead.

    ReplyDelete
  3. Why is chess interesting?

    Even if we restrict ourselves to abstract turn-based strategy games, why does chess stand out? What motivates its rules?

    This is a serious question. I would like to make up my own abstract computer-based turn-based strategy games. So any insight here would be appreciated.

    ReplyDelete
  4. :: What makes chess interesting?

    An excellent question, and it leads into matters I've taken seriously but haven't gathered the p(hysi/sych)ological background to analyze. Here are some elements I point to---one can compare/contrast to the game of Go for most of them.

    (1) Immediate "Gestalt" visual appeal: There are several different kinds of pieces, each with a "personality", and not too many pieces on a fairly small board.

    (2) Individual tactics usually have a fairly short horizon.

    (3) The Pawns lend "structure" to the game. Combined with (2), it is often possible to reach a basic understanding of a position in a short amount of time. I.e. the game "reads well".

    (4) There is a buffer between winning and losing. I wish chess could be less "drawish"---e.g. I wonder about the effect the Arabian rules, which allow win by stalemate and by playing one's King to the center when the opponent is left with just the King. But it is nice that the boundary between win and defeat is not razor sharp.

    (5) Individual games can be played over and studied in a relatively short time.

    (6) Not only does chess have a long history, the game has developed a vocabulary that allows strategical and tactical ideas of games to be conveyed quickly and intuitively.

    One effect of (1)--(3), however, is that chess is not very "deep", in terms that can be well-defined as follows: Say Player Y is "one class above" Player X if Y wins 75% of games between them. The Elo Rating System is usually configured so that "one class above" == a difference of 200 rating points. In the US, a rating of 2400+ like mine is called "Senior Master", 2200 is "Master", 2000 is "Expert", 1800 is "Class A", 1600 is "Class B", and so on. Absolute beginners come in about 600-700, while Kasparov maxed out at 2851 (international scale, maybe = about 2900 US). So the depth of chess is about 11 "class units".

    Rankings in Go and the handicap system used in Go do not make this concept easy to carry over, but in the mid-90s I read estimates of 25--40 for the depth of Go. This doesn't mean Go is 2-4x as deep as Chess, but rather thousands if not millions of times deeper, as it's an exponential scale! And in my opinion, this notion of depth is the lone main predictor of computer vs. human strength in a game, keeping processing speeds and other stuff equal.

    The Rybka program has been rated at 3122 running on a quad-core 64-bit PC similar to the one on which Deep Fritz 10 beat Kramnik in a match Nov-Dec. 2006 (the one where Kramnik stepped into a mate-in-one, but DF10's other win is the strongest I've ever seen one side play, truly "3,000+ chess"). Thus programs are getting 2-or-more class units ahead of humans, and since the tweaks I can think of for chess would add 1 class unit at most to its depth, I think humans are cooked at chess. If I recall correctly, Japanese chess (Shogi) has a depth of 14, so by my prediction, Shogi engines optimized on the same hardware should be able to take 25%-50% against top human players. The Wikipedia article I referenced talks about only 1 top-level human-computer Shogi game being played, in March of last year, a 112-move win for the human but after the computer missed a chance to win.

    The game Arimaa was expressly designed to be deeper than chess, while using the same-size board and armies and retaining the flavor of human appeal of my points (1)--(6). Humans last spring convincingly beat Arimaa programs that were adapted from chess engines, and whether one can design an engine tuned specifically for Arimaa that can beat humans is a $10,000+ challenge on their site. This can be your first comparison point in developing a new game of strategy. All the best!

    ReplyDelete
  5. Baseline chess?? Are you kidding me?? Pal Benko proposed this in the 1970's and personally told me Fischer's version was "of course" a ripoff of his "PRECHESS". He also said he didn't want any trouble with Bobby and he wasn't making any money off it. Also the so called Fischer clock was in use for Shogi many years before he stole that idea also. Why didn't anyone else thing of that?? For both these examples cited, someone else did!!

    ReplyDelete