|
|
Comparison Between Chess And Go
© 2008 Milton N. Bradley
Although manifestly quite different, Chess and Go are unquestionably the two finest strategic board games in all of human history. Some prefer one, some the other, and there are a small fortunate number who enjoy and excel at both. So why compare these two exemplars in a manner designed to demonstrate that one is superior to the other? The answer to that important question is at least partially provided by the lead editorial in The New York Times of May 10, 1997 entitled "Mind Over Matter", which commented on the then ongoing Chess match between IBM's Deep Blue Computer Program and World Chess Champion Gary Kasparov, in part as follows:
"No one much cared when computers mastered backgammon and checkers or clobbered lesser Grandmasters in chess. But now that we have sent the greatest chess champion in human history into battle, the prospect of defeat seems unnerving. Still, before mere mortals sink too deeply into despair, it is important to recognize several comforting alibis that may apply here.
Deep Blue is not thinking the way humans do. It is using its immense number-crunching power to explore millions of moves per second, and applying a set of rules provided by its human masters to pick the strongest. This gives it tremendous powers to play chess, a narrow, circumscribed pursuit that is red meat for high-speed computation but hardly the supreme measure of intelligence its practitioners like to pretend."
Deep Blue won its match with human World Chess Champion Kasparov by causing his resignation in Game 6 after only 19 moves, after what had been an even contest to that point. The following is a typical reaction to this event as posted on the web newsgroup rec.games.chess.misc: "Deep Blue won that final game not because of any superiority it had over Kasparov, but because of bizarre opening play by Kasparov.". To my mind that's a sure indication that Kasparov "cracked" psychologically, but whether or not that alone is sufficient to justify characterizing his defeat as "decisive" I leave to the reader's judgment.
The key thing that's almost certain is that with the inevitable improvements that might have been readily added to its hardware and software in subsequent months had IBM chosen to continue its development, Deep Blue would likely have become indisputably superior. But having satisfied whatever corporate objectives they had, once having achieved that narrow victory IBM "pulled the plug" and disassembled Deep Blue, presumably forever.
Now it appears that those improvements have come to pass, as explained in the Chess column of The New York Times on Sunday Feb 17, 2008, when Grandmaster Joel Benjamin wrote: 揑f a computer played perfectly, could any human hope to achieve a draw? Chess engines are still far from perfection, but their advances have been staggering. Deep Blue made history in 1997 by defeating Gary Kasparov in a six-game match. A decade later, no human would dare take on a chess program at even strength. The premier chess engine, Rybka, is estimated at (Elo) 3100, or 300 points higher than any player. Rybka is exploring this dominant relationship with handicap matches against grandmasters. After narrowly losing a match at pawn handicap (the rough equivalent of 2 stones in Go? MB) I agreed to participate in an experiment. Every draw would count as a win for me. After we split 4 games, Rybka streaked to a 6-2 rout. Rybka played Black in all 8 games. Larry Kaufman, co-programmer of Rybka, opined that no human has a chance without first move. Playing White greatly increases the chances for a grandmaster to draw with Rybka....?
In sharp contrast, the best computer Go programs are still mired at just beyond an advanced beginner's level, despite the presence for over 15 years of a $1 million prize for a program which can defeat a master Go professional, offered by the Ing Chang-Ki Goe (his unique spelling!) Educational Foundation of Taiwan. But no claimants for this impressive prize are even visible on the far horizon after all these years! The many reasons for this disparity are set out in the comparison below, but the most obvious concerns the many orders of magnitude vastly greater size of the "move trees" in Go, which render the massive number-crunching power of Deep Blue and even its potentially vastly more powerful descendants totally impotent in the critical opening and middle phases of the game.
At least equally important is the fact that, unlike Go, Chess essentially lacks a deep strategic component! To those who may question this assertion, substantiation is provided by the following famous quote:
"Chess is 99% tactics." -- Richard Teichmann, (1868-1925), a German Grandmaster who for many years lived in England, and one of the strongest attacking players of all time.
Al Lawrence, former Executive Director of the US Chess Federation said of this quote: "Everyone has always agreed on this point--even before Teichmann, who happened to come up with a snappy way to say it."
This was validated by Grandmaster Reuben Fine in his book "Chess Marches On", published by Chess Review, 1945, in which he said on page 97:
"Thirty years ago Teichmann said that chess is 99% tactics. And despite the enormous strides of chess theory since then, his percentage can only be reduced (by) a few points. Many amateurs think that master games are usually decided by some deeply-laid plan covering all possibilities for at least ten moves.. That is what they conceive the grand strategy of tournaments to be. Actually, however, strategical considerations, while quite important, do not cover a range or depth at all comparable to the popular notion. Very often, in fact, sound strategy can dispense with seeing ahead at all, except in a negative or trivial sense. And it is still true that most games, even between the greatest of the great, are decided by tactics or combinations which have little or nothing to do with the fundamental structure of the game."
Then there is this excerpt from the internet review "NEW CHESSBASE CDs", by Steve Lopez, who said:
"It's a hoary old cliché that's been kicked around for three-quarters of a century. But do you know why it's quoted so often? Because Teichmann was right! If Philidor was right in saying that the pawn is the soul of chess, Teichmann was correct in implying that tactics is the heart of the game."
Some of my master level chess-playing friends have complained that by making this comparison and demonstrating the objective superiority of Go I am denigrating Chess, but this is a misperception. For example, if we compare precious metals and conclude that gold is more valuable than silver, are we denigrating silver? Or just acknowledging objective reality? It's clear that silver has many vital uses as coinage, an electrical conductor, and in jewelry, so it is of importance in industry, commerce and quality of life areas. Under these circumstances our objective appraisal of its position as less valuable than gold should offend no unbiased observer. And the same should hold true for Chess compared with Go, but regrettably it almost invariably does not. Chess is interesting in its own right and has many assets, but when all of its attributes are stacked up against those of Go it necessarily ranks second best! The problem is that most chessplayers refuse to even contemplate that this could be true!
What most chessplayers share in this regard is that they are so convinced of the superiority of their game that, almost invariably with at best meager knowledge of Go gained almost entirely by hearsay, they dismiss the entire idea without even exhibiting a willingness to find out for themselves. This is the very problem that Chessmaster Edward Lasker and then World Chess Champion Emanuel Lasker encountered in 1905 in Germany, and which is described in the prior section of this web page entitled "How Go Came To America". Maintaining this close-minded posture is any chessplayer's absolute right, of course, but its sad consequence is that it unnecessarily cuts them off from one of life's greatest purely intellectual challenges and pleasures. Perhaps an even worse loss as a result of their narrow perspective, those chessplayers of my acquaintance who have put forth the effort to study Go report that it has improved their chess, so this is yet another detriment that those who display this "head-in-the-sand" attitude sustain. Hopefully those of you who are reading this do not wear those same blinders, and are willing to investigate Go further and then judge for yourselves!
Finally, there is this perhaps most cogent comment on the comparison between Chess and Go that appeared in an article in The New York Times Metro Section of Thursday February 6, 2003 entitled "Queen, Captured by Mouse", which focused on the then ongoing tied match between Gary Kasparov and the Israeli Chess software program "Deep Junior". That article featured the cogent quote by Dr. Hans Berliner that appears below, whose relevance can only be fully appreciated if you understand Berliner's background and outstanding credentials in the field. These were spelled out in NY Times Bridge columnist Phillip Alder's column on Saturday, Nov 4, 2006 as follows: "Hans Berliner, who was well known in chess circles, used to be a top bridge player, but retired from that game some 50 years ago. Berliner won the 1956 Eastern States open Chess Championship ahead of Bobby Fischer. But he gave up tournament chess to become the world's leading correspondence player, winning that world championship, and building a chess-playing computer. During the early stages of writing the chess program, Berliner realized that not enough was known about positional evaluation. So he turned to Backgammon. The result was BKG 9.8, the first computer program to beat a World Champion in any game when it won a backgammon match against Luigi Villa in June 1979. This research led, in 1984, to a chess program called HiTech. Berliner went on to become a Professor of Computer science at Carnegie Mellon University, from which he retired in 1998 at 69." Because Berliner's work on chess led to the development of IBM's Deep Blue and its descendants, it had great impact and relevance when he said: "You don't have to be really good anymore to get good results. Chess is winding down.....What's happening with Chess is that it's gradually losing its place as the par excellence of intellectual activity". And he concluded: "Smart people in search of a challenging board game might try a game called Go..."
Need I really say more?
Comparison Between Chess And Go
CRITERION CHESS GO
When Invented Reputedly About 455 A.D., Probably Much Earlier. About 2000 B.C.
Where Invented Reputedly India, Probably China / China
Suitable For Players 3 years and older / 3 years and older
Number of Players 2, White and Black 2, / Black and White
Color Selection Chosen by Lot The weaker player always takes Black. /Equal players alternate colors in successive games.
First Play (= The Initiative) Always White Always Black, except that in handicap games / Black's handicap stones count as his first move so that White actually makes the first "free" play.
Rules of Play Complex: Different piece moves, many special rules (e.g. en passant capture, castling, etc.) / Elegant: Only 2 simple rules govern all play.
Scope/Board Size Restricted: Only 8 x 8 = 64 squares. / Immense: 19 x 19 = 361 intersections.
Object Of The Game Checkmate Opposing King = Total Victory / Obtain Larger Territory = Greater "market share"
Brain Functions Used In Playing (1) Almost Entirely Analytical (left brain). / Fully utilizes/integrates analytic (left brain) and artistic/pattern recognition (right brain) functions.
Number/kind of Playing Pieces 1 King, 1 Queen, 2 ea. Rooks, Knights, & Bishops + 8 Pawns/side. / An unlimited supply of uniform double-convex lens shaped pieces (called stones) per side.
The Moves Of The Pieces Each type of piece has its own unique move. None. Stones are simply entered on the board one-per-turn on any empty intersection according to the rules of Go, and then are never moved thereafter unless captured (when they are removed from the board).
The Starting Lineup. Fixed by custom. None. The board begins empty and the players enter their stones anywhere that they wish to uniquely structure the board in every game.
Captures Made one-at-a-time, by moving the capturing piece into the square occupied by the piece being captured. Captured singly or en masse, by surrounding the captive(s) so that they are not connected to any adjacent open intersection.
The Fate Of Captives. Removed from the board (and further play). Removed from the board and held as prisoners, which are subtracted from the opponent's score at game's end. Each prisoner is worth one point.
Number of possible First Moves. 20 White x 20 Black = 400. 361 Black x 360 White = 129960, although symmetry reduces this number to an effective 32,490.
Number of STRONG First Moves 8 White x 8 Black = 64 32 Black x 31 White = 992
The Value Of First Move Not precisely evaluated, but often considered sufficient to force a draw and close to enough to confer a winning advantage. Most often evaluated at 6 1/2 points (called Komi), which are added to White's final score to compensate for playing second. The extra 1/2 point prevents draws.
Estimated Number of Possible Board Configurations 10 120 OMNI Magazine in June, 1991 proposed 10 761, but most believe that the correct figure is really on the order of 10 174.
Opening "Book" Structures the entire board. Any given "book" line is always feasible. Structures only a single corner. Integration of the 4 corners with the sides and center is governed by general strategic principles and tactical constraints. The feasibility of any given corner sequence depends upon what has already happened as well as the player's strategic plan for the remainder of the game.
Handicap System No formal system. Giving handicaps of one or more pieces or pawns imprecisely compensates for differences in playing strength, and distorts both tactics and strategy. Handicaps consist of allowing Black to place an appropriate number of stones on the board before White's first move. This natural and very precise compensation for playing strength differences is completely consistent with normal play.
Military Analogy A single battle. An entire multi-front war.
The Nature of Play Primarily tactical, with only a modest strategic component. Profoundly strategic, but with incisive, complex, integral tactics.
The Nature Of The Opening Primarily a struggle for development, center control, and maneuvering space, with subsidiary objectives of pawn structure, king safety, and material. Each side establishes a series of delicate balances/tradeoffs between the conflicting yet complementary objectives of territorial acquisition, "influence", making "shape", maintaining center access, and attack and defense.
The Nature Of The Middle Game Mainly tactical piece maneuvering to attempt to attain a winning material advantage or to mount a successful mating attack, often involving real and/or pseudo sacrifices, but also with strategic elements involving piece mobility, center control and pawn structure. Features deep strategic planning implemented via incisive tactics involving feints, diversions, invasions, pincer and multi-purpose attacks,, often involving both real and pseudo sacrifices.
The Nature Of The Endgame Piece and pawn maneuvers to force and/or consolidate a winning material advantage, promote a pawn, or force mate. "Endgame" moves always occur at the physical end of the game. Final consolidation of territorial borders between safe opposing armies. Although these sequences typically involve only a few points each, maintenance of the initiative throughout a sequence of such plays can readily result in the gain of enough points to decide the game. Because of the immense board size, local "endgame" sequences frequently occur at almost every stage of the game.
How Victory Is Decided. (2) Checkmate, resignation, or time. Draws are possible by agreement, stalemate, repetition of position, or perpetual check. Higher final score, resignation, or time. No draws are possible because repetition of any full board position is prohibited, and because of the extra half point given to White as part of his compensation for Black's first move advantage.
Business/Strategic Planning Analogy Emphasis on short term profits. Minimal attention to automation, Quality Control, and investment in capital equipment/plant modernization. Emphasis on long term planning, including automation, Quality control, and investment in capital equipment/plant modernization.
Countries Using This Kind Of Thinking In Their Business/Political Decision Making. US, Western Democracies, Russia, and Eastern European Nations. Japan, Taiwan, South Korea, Hong Kong, Singapore.
Strength Of computer Programs Grandmaster Level Intermediate (5-8 Kyu) level
Reasons For Computer Program Performance Restricted scope makes application of "standard" techniques such as heuristic search, etc. feasible even on microcomputers. Immense scale makes the application of "standard" techniques infeasible even on supercomputers. Requires a real breakthrough in Artificial Intelligence which has not yet been achieved.
Abstract
GO is a board game thought to be different from chess in many aspects, most significantly in that GO emphasizes global strategy more than local battle, a property very difficult for computer programs to emulate. To investigate the neural basis of GO, functional magnetic resonance imaging (fMRI) was used to measure brain activities of subjects engaged in playing GO. Enhanced activations were observed in many cortical areas, such as dorsal prefrontal, parietal, occipital, posterior temporal, and primary somatosensory and motor areas.
Quantitative analysis indicated a modest degree of stronger activation in right parietal area than in left. This type of right hemisphere lateralization differs from the modest left hemisphere lateralization observed during chess playing.
(2) At first blush, the idea of a draw is superficially appealing! After all, if two opponents are really almost exactly equal in strength, why should the result of their encounter be that one gets a full point and the other gets nothing? Especially if no obviously egregious errors have been made, it seems fairer that they split the point evenly. But upon deeper introspection, that idea begins to fall apart. And this is something that most chessplayers implicitly acknowledge, but then act as though they were either oblivious to or deliberately choose to ignore as less than significant. That this attitude is at best mistaken was made abundantly clear by Robert Byrne, the distinguished New York Times chess columnist, in his regular weekly column of Sunday, March 28, 2004 when he said:
l adimir Kramnik of Russia, the world champion, won the Super GM Tournament in Linares, Spain, held from Feb. 19 to March 5. He scored 7-5 in the elite seven-entrant double-round-robin invitational event. Peter Leko and Garry Kasparov tied for second with 61/2 -51/2. It is the greatest satisfaction in the game to be No.1 in a tourney where everyone is at the top of the world rankings. That having been said, this competition had only nine decisive encounters. All the rest were draws.?
Translating Byrne抯 statement a bit to make clearer the point he抯 making, we see that of a total of 84 games contested a full 89% resulted in a draw! And how many of those were actually infamous 揼randmaster draws? in which the opponents, wary of each other抯 skills and unwilling to risk a loss, really only 揼o through the motions? for a few desultory moves and then quickly agree to a face saving draw, isn抰 revealed. The fact that Byrne even explicitly pointed this out is, I believe, ample proof that even among the most ardent chess lovers this remains a serious intrinsic flaw that greatly diminishes both the game抯 integrity and the satisfaction its players derive from it.
-------------------------------------------------------------------------------- |
|
发表于 2010-5-31 08:04:27
楼主
匿名
发表于 2010-6-2 08:02:22