About Opentopia
Opentopia Directory Encyclopedia Tools

Computer chess

Encyclopedia : C : CO : COM : Computer chess

1990s Pressure-sensory Chess Computer with LCD screen
Enlarge
1990s Pressure-sensory Chess Computer with LCD screen

The idea of creating a chess-playing machine dates back to the eighteenth century. Around 1769, the chess playing automaton called The Turk became famous before being exposed as a hoax. After that, the field of mechanical chess research languished until the advent of the digital computer in the 1950s. Since then, chess enthusiasts and computer engineers have built, with increasing degrees of seriousness and success, chess-playing machines and computer programs.

Chess-playing computers are available for negligible cost, and there are many programs (such as free software, like GNU Chess, Amy and Crafty) that play a game that, with the aid of virtually any modern personal computer can defeat most master players under tournament conditions, while top commercial programs like Shredder or Fritz have surpassed even world champion caliber players at blitz and short time controls.

Chessmaster 10th edition running on Windows XP
Enlarge
Chessmaster 10th edition running on Windows XP

Fritz 6 running on Windows XP
Enlarge
Fritz 6 running on Windows XP

Rebel Tiger II running in player versus computer mode on Windows XP
Enlarge
Rebel Tiger II running in player versus computer mode on Windows XP

Contents

Motivation

The prime motivations for computerized chess playing have been solo entertainment (allowing players to practice and to amuse themselves when no human opponents are available), as aids to chess analysis, for computer chess competitions, and as research to provide insights into human cognition. For the first two purposes computer chess has been a phenomenal success - going from the earliest real attempts to programs which challenge the best human players took less than fifty years. We can say that chess play is not an intractable problem to modern computing.

However, to the surprise and disappointment of many, chess has taught us little about building machines that offer human-like intelligence, or indeed do anything except play excellent chess. For this reason, computer chess, (as with other games, like scrabble) is no longer of great academic interest to researchers in artificial intelligence, and has largely been replaced by more intuitive games such as Go as a testing paradigm. Chess-playing programs essentially explore huge numbers of potential future moves by both players and apply a relatively simple evaluation function to the positions that result, whereas computer Go challenges programmers to consider conceptual approaches to play.

Brute-force methods are useless for most other problems artificial intelligence researchers have tackled, and are very different from how human chess players select their moves. In some strategy games, computers easily win every game, while in others they are regularly beaten even by amateurs. In chess, the combined skills of knowledgeable humans and computer chess engines can produce a result stronger than either alone.

3-D board, Fritz 8
Enlarge
3-D board, Fritz 8

Brute force versus strategy

The first paper on the subject by Claude Shannon, published in 1950 before anyone had programmed a computer to play chess, successfully predicted the two main possible search strategies which would be used, which he labeled 'Type A' and 'Type B'.

Type A programs would use a "brute force search" approach, examining every possible position for a fixed number of moves using the minimax algorithm. Shannon believed this would be impractical for two reasons.

Firstly, with approximately 30 moves possible in a typical real-life position, he expected that searching the 30 × 30 × 30 × 30 × 30 × 30 positions involved in looking three moves ahead for both side (six plies) would take about 16 minutes, even in the "very optimistic" case that the program evaluated a million positions every second. (In the event, it took about forty years to achieve this speed.)

Secondly, it ignored the problem of quiescence, trying to only evaluate a position that is at the end of an exchange of pieces or other important sequence of moves ('lines'). He expected that adapting type A to cope with this would greatly increase the number of positions needing to be looked at and slow the program down still further.

Claude Shannon
Enlarge
Claude Shannon

Instead of wasting processing power examining bad or trivial moves, Shannon suggested that type B programs would use a "strategic AI" approach to solve this problem by only looking at a few good moves for each position. This would enable them to look further ahead ('deeper') at the most significant lines in a reasonable time.

Adriaan de Groot interviewed a number of chess players of varying strengths, and concluded that both masters and beginners look at around forty to fifty positions before deciding which move to play. What makes the former much better players is that they use pattern recognition skills built from experience. This enables them to examine some lines in much greater depth than others by simply not considering moves they can assume to be poor.

More evidence for this being the case is the way that good human players find it much easier to recall positions from genuine chess games, breaking them down into a small number of recognisable sub-positions, than completely random arrangements of the same pieces. In contrast, poor players have the same level of recall for both.

The problem with type B is that it relies on the program being able to decide which moves are good enough to be worthy of consideration ('plausible') in any given position and this proved to be a much harder problem to solve than speeding up type A searches with superior hardware.

One milestone was the abandonment of type B in 1973 by the team from Northwestern University responsible for the 'Chess' series of programs, who had won the first three ACM Computer Chess Championships (1970-1972). The resulting type A program, Chess 4.0, won that year's championship and its successors went on to come second in both the 1974 ACM Championship and that year's inaugural World Computer Chess Championship, before winning the ACM Championship again in 1975, 1976 and 1977.

One reason they gave for the switch was that they found it less stressful during competition, because it was difficult to anticipate which moves their type B programs would play and why! They also reported that type A was much easier to debug in the four months they had available and turned out to be just as fast: in the time it used to take to decide which moves were worthy of being searched, it was possible to just search the lot.

Temporarily, the brute force camp won out for these reasons and in the sense that their programs simply played better chess. The game of chess is not conducive to inerrantly discriminating between obviously bad, trivial, and good moves using a rigid set of rules. Traps are set and sprung by expert players who understand and master the many levels of depth and irony inherent to the game.

Furthermore, technological advances by orders of magnitude in processing power have made the brute force approach far more incisive in the 90's than was the case in the early years. The result is that a very solid, tactical AI player has been built which is errorless to the limits of its search depth and time. It turned out to produce excellent results, at least in the field of chess, to let computers do what they do best (calculate) rather than coax them into imitating human thought processes and knowledge. In 1997 Deep Blue, a Type A machine, defeated World Champion Garry Kasparov, marking the first time a computer has defeated a reigning world chess champion in standard time control.

However, starting in the late 1990s Type B programs began replacing Type A as the preferred method for programmers. In 1998, Rebel 10, a commercially available Type B program, defeated Viswanathan Anand who at the time was ranked second in the world, by a score of 5-3. But one should note that out of the four blitz games (5 minutes + 5 seconds Fischer for each move), Rebel won 3-1, in the two semi-blitz games (15 minutes) Rebel won 1.5 -0.5 and in the tournament game (40/2:00, 1 hour), it was Anand who won. Noteworthy is that computers play better at faster time controls, but a player's actual strength is only determined at the classical time controls, where Anand proved that Humans were still better[link]. In the early 2000's, commercially available Type B programs such as Deep Junior and Fritz were able to draw matches against former world champion Garry Kasparov and former "classical" world champion Vladimir Kramnik. In 2005, Hydra, a Type B chess machine, defeated Michael Adams who at the time was ranked 7th in the world, by an overwhelming score of 5.5-0.5.

Computers versus humans

On the other hand, it remained an open question whether any amount of brute force computation would ever be adequate to defeat the expertise of top humans. In 1968, IM David Levy made a famous bet that no chess computer would be able to beat him within ten years. He won his bet in 1978 by beating Chess 4.7 (the strongest computer at the time), but acknowledged then that it would not be long before he would be surpassed. In 1989, Levy was crushed by the computer Deep Thought in an exhibition match.
Kasparov vs. Deep Blue
Enlarge
Kasparov vs. Deep Blue

Deep Thought, however, was still considerably below World Championship Level, as the then reigning world champion Garry Kasparov demonstrated in two sterling wins in 1989. It wasn't until a 1996 match with IBM's Deep Blue that Kasparov lost his first game to a computer at tournament time controls in Deep Blue - Kasparov, 1996, Game 1. This game was, in fact, the first time a reigning world champion had lost to a computer using regular time controls. However, Kasparov regrouped to win three and draw two of the remaining five games of the match, for a convincing victory.

In May 1997, an updated version of Deep Blue defeated Kasparov 3.5-2.5 in a return match. IBM keeps a web site of [the event]. While not an official world championship, the outcome of the match is often taken to mean that the strongest player in the world is a computer. Such a claim is open to strong debate, as a truly fair human-machine match is difficult to arrange.

tright

IBM retired Deep Blue after the match and it has not played since. However, other "Man vs. Machine" matches continue to be played. In October 2002, Vladimir Kramnik and Deep Fritz competed in the eight-game Brains in Bahrain match, which ended in a draw. Kramnik won games 2 and 3 by "conventional" anti-computer tactics—play conservatively for a long-term advantage the computer is not able to see in its game tree search. Fritz, however, won game 5 after a severe blunder by Kramnik. Game 6 was described by the tournament commentators as "spectacular." Kramnik, in a better position in the early middlegame, tried a piece sacrifice to achieve a strong tactical attack, a strategy known to be highly risky against computers who are at their strongest defending such attacks. True to form, Fritz found a watertight defence and Kramnik's attack petered out leaving him in a bad position. Kramnik resigned the game, believing the position lost. However, post-game human and computer analysis has shown that the Fritz program was unlikely to have been able to force a win and Kramnik effectively sacrificed a drawn position. The final two games were draws. Given the circumstances, most commentators still rate Kramnik the stronger player in the match.

Grandmaster Michael Adams
Enlarge
Grandmaster Michael Adams
Hydra, a dedicated chess computer with custom hardware and sixty-four processors, has defeated other top computer programs in recent matches, and is probably the strongest computer player yet created.  It crushed seventh-ranked Michael Adams 5.5-0.5 in a recent six-game match (though Adams' preparation was far less thorough than Kramnik's 2002 series).  Some commentators [link] believe that Hydra will ultimately prove clearly superior to the very best human players, or if not its direct successor will.  


Several other matches with computers against top human players were played recently:



Endgame databases


Computers have been used to analyse some chess endgame positions completely. Such endgame databases are generated in advance using a form of retrograde analysis, starting with positions where the final result is known (e.g. where one side has been mated) and seeing which other positions are one move away from them, then which are one move from those etc. Ken Thompson, perhaps better known as the key designer of the UNIX operating system, was a pioneer in this area. 


Endgame play had long been one of the great weaknesses of chess programs because of the depth of search needed, with some otherwise master-level programs being unable to win in positions that even intermediate human players would be able to force a win.


The results of the computer analysis sometimes surprised people. Thompson's Belle chess machine proved itself able to draw the theoretically lost ending of King and Rook against King and Queen despite not following the usual strategy to delay defeat of keeping the two pieces close together. Asked to explain the reasons behind some of the program's moves, Thompson was unable to do so beyond saying the program's database simply said the moves were the best.


Other positions, long believed to be 'won', turned out to take more moves against perfect play to actually win than were allowed by chess's fifty move rule. As a consequence, for some years the official laws of chess were changed to extend the number of moves allowed in these endings. After a while, the law reverted back to fifty moves in all positions - more such positions were discovered, complicating the rule still further, and it made no difference in human play, as they could not play the positions perfectly.


Over the years, other endgame database formats have been released including the Edward Tablebases, the De Koning Endgame Database (released in 2002) and the Nalimov Endgame Tablebases which is the current standard supported by most chess programs such as Shredder or Fritz. All five-piece endings have now been analysed completely, and some of the six-piece endings are available.


The databases are generated by storing in memory the values of positions which have been encountered so far, and using these results to lop off the ends of the search trees if they arise again. Although the number of possible games after a number of moves rises exponentially with the number of moves, the number of possible positions with a few pieces is exponential only in the number of pieces - and effectively limited however many end game moves are searched. The simple expediency of remembering the value of all previously reached positions means that the limiting factor in solving end games is simply the amount of memory available in the computer. While computer memory sizes are increasing exponentially, there is no reason why end games of increasing complexity should not continue to be solved.


A computer using these databases will, upon reaching a position in them, be able to play perfectly, and immediately determine whether the position is a win, loss or draw, plus the fastest way of getting to that result. Knowledge of whether a position is a win, loss or draw is also helpful in advance since this can help the computer avoid or head towards such positions depending on the situation. 


Endgame databases featured prominently in 1999, when Kasparov played an exhibition match on the Internet against the Rest of the World. A seven piece Queen and pawn endgame was reached with the World Team fighting to salvage a draw. Eugene Nalimov helped by generating the six piece ending tablebase where both sides had two Queens which was used heavily to aid analysis by both sides.



Computer chess implementation issues

Developers of chess-playing computer system must decide on a number of fundamental implementation issues. These include

Implementors also need to decide if they will use endgame databases or other optimizations, and often implement common de facto chess standards.



Board representations


The data structure used to represent each chess position is key to the performance of the application.  Methods range from obvious and easy (an array) to obscure and difficult Bitboard.



Search techniques

Computer chess programs consider chess moves as a game tree. In theory, they examine all moves, then all counter-moves to those moves, then all moves countering them, and so on, where each individual move by one player is called a "ply". This evaluation continues until it reaches a final "leaf" position which is evaluated.


A naïve implementation of this approach, however, would never finish in a practical amount of time, so various methods have been devised to greatly speed the search for good moves.


For more information, see:





Leaf evaluation

For most chess positions, computers cannot look ahead to all final possible positions. Instead, they must look ahead a few ply and then evaluate the final board position. The algorithm that evaluates final board positions is termed the "evaluation function", and these algorithms are often vastly different between different chess programs.


Evaluation functions typically evaluate positions in hundredths of a pawn, and consider material value along with other factors affecting the strength of each side. When counting up the material for each side, typical values for pieces are 1 point for a pawn, 3 points for a knight or bishop, 5 points for a rook, and 9 points for a queen. (See chess piece point value.)  By convention, a positive evaluation favors White, and a negative evaluation favors Black.


The king is sometimes given an arbitrary high value such as 200 points to ensure that a checkmate outweighs all other factors. Evaluation functions take many factors into account, such as pawn structure, the fact that a pair of bishops are usually worth more, centralized pieces are worth more, and so on.  The protection of kings is usually considered, as well as the phase of the game (opening, middle or endgame).


See Claude Elwood Shannon for a description of his early paper about a chess-playing program.



Using endgame databases


Some computer chess operators have suggested that endgame tablebases will actually weaken performance strength in chess computers. Because some positions are analyzed as forced wins for one side, the program will avoid the losing side of positions at all costs. However, many endgames are forced wins only with flawless play, where an even slight error would produce a different result. Consequently, most modern engines will play many endgames well enough on their own. A symptom of this problem is that computers may resign too early because they see that they are being forced into a position that is theoretically dead lost (although they may be thirty or more moves away from the end of the game, and most human opponents would find it hard to win in that time). 


The Nalimov tablebases do not consider the fifty move rule, under which a game where fifty moves pass without a capture or pawn move can be claimed to be a draw by either player. This results in the tablebase returning results such as "Forced mate in 66 moves" in some positions which are actually a dead draw because of the fifty move rule. 


One reason for this is that if the rules of chess were to be changed once more, giving more time to win such positions, it will not be necessary to regenerate all the tablebases. It is also very easy for the program using the tablebases to notice and take account of this 'feature'. 


The Nalimov tablebases, which use state-of-the-art compression techniques, require 7.05 GB of hard disk space for all five-piece endings. It is estimated that to cover all the six-piece endings will require at least 1 terabyte. It is estimated that seven-piece tablebases will require more storage capacity than will be available in the foreseeable future. 


Since endgame positions are typically very simple, and with the power of desktop computers growing exponentially, most engines can calculate in endgames very effectively, making the usefulness of endgame databases questionable in light of these serious shortcomings.



Other optimizations

Many other optimizations can be used to make chess-playing programs stronger. For example, transposition tables are used to record positions that have been previously evaluated, to save recalculation of them. Refutation tables record key moves that "refute" what appears to be a good move; these are typically tried first in variant positions (since a move that refutes one position is likely to refute another). Opening books aid computer programs by giving common openings that are considered good play (and good ways to counter poor openings).


Of course, faster hardware and additional processors can improve chess-playing program abilities, and some systems (such as Deep Blue) use specialized chess hardware instead of solely software implementations.



Standards

Computer chess programs usually support a number of common de facto standards. Nearly all of today's programs can read and write game moves as Portable Game Notation (PGN), and can read and write individual positions as Forsyth-Edwards Notation (FEN). Older chess programs often only understood long algebraic notation, but today users expect chess programs to understand standard algebraic chess notation.


Most computer chess programs are divided into an engine (which computes the best move given a current position) and a user interface. Most engines are separate programs from the user interface, and the two parts communicate to each other using a public communication protocol. The most popular protocol is the Xboard/Winboard Communication protocol. Another open alternate chess communication protocol is the Universal Chess Interface. By dividing chess programs into these two pieces, developers can write only the user interface, or only the engine, without needing to write both parts of the program. (See also List of chess engines.)



Playing strength versus computer speed

It has been estimated that doubling the computer speed gains approximately 50-70 ELO points in playing strength. However, this applies mainly to computer-vs-computer matches, and not to computer-vs-human matches.



Other chess software

There are several other forms of chess-related computer software, including the following:


Computer chess engine tournaments



Chess engine rating lists



Advanced chess


Advanced Chess is a form of chess developed in 1998 by Kasparov where a human plays against another human, and both have access to computers to enhance their strength. The resulting "advanced" player was argued by Kasparov to be stronger than a human or computer alone, although this has not been proven.



Computer chess theorists

Well-known computer chess theorists include:


The future of computer chess?

Many observers extrapolate that computers will consistently beat the best human players by perhaps 2010, and then go on to exceed their abilities to the point where a human versus computer chess match would be as unfair as a human versus automobile race. Others are unconvinced, saying that there are still deep strategic elements to chess that will resist brute-force computer searching.


If chess computers become too big of a match for humans, the field of computer versus computer competition (where makers of chess computers pit their best machines against each other) will take over and it already has its start.


One potentially fruitful field of research is in distributed computation, where many computers are joined together through the internet and are each tasked with a small section of the overall search tree to analyse. The leading project is the [ChessBrain project], which gained a world record in 2004 for the largest number of computers ever playing a game of chess simultaneously (2,070).



Solving chess

Some two-player board games have been solved; that is, the "perfect" strategy has been devised.  As the number of possible chess positions is finite, a perfect strategy for chess must exist.  However, determining this strategy, and the resulting outcome, is currently completely computationally infeasible and will likely remain so without fundamental breakthroughs in computer technology.  


The number of board positions that a chess board may be placed in is astronomical—more than the number of atoms in the universe. However, the number of legal positions is rather smaller, under  1050. This is somewhat too large to be solved by conventional technology in less than a few centuries, even with a thousand fast chess computers each searching a billion positions per second, and using alpha-beta pruning.



Chronology of computer chess





tright







 

-->


 

-->































































































































 

-->


 

-->





 










See also



External links



















 





From Wikipedia, the Free Encyclopedia. Original article here. Support Wikipedia by contributing or donating.
All text is available under the terms of the GNU Free Documentation License See Wikipedia Copyrights for details.














Search Titles









0123456789
ABCDEFGHIJ
KLMNOPQRST
UVWXYZ?










E-mail this article to:



Personal Message: