Deep Blue was[a] asupercomputer forchess-playing based on a customizedIBM RS/6000 SP. It was the first computer to win a game, and the first to win a match, against a reigning world champion under regular time controls. Development began in 1985 atCarnegie Mellon University under the nameChipTest. It then moved to IBM, where it was first renamedDeep Thought, then again in 1989 to Deep Blue. It first played world championGarry Kasparov in asix-game match in 1996, where it won one, drew two and lost three games. It was upgraded in 1997, and in a six-game re-match it defeated Kasparov by winning two games and drawing three. Deep Blue's victory is considered a milestone in thehistory of artificial intelligence and has been the subject of several books and films.
After Deep Thought's two-game 1989 loss to Kasparov, IBM held a contest to rename the chess machine: the winning name was "Deep Blue", submitted byPeter Fitzhugh Brown,[8] was a play on IBM's nickname, "Big Blue".[b] After a scaled-down version of Deep Blue played GrandmasterJoel Benjamin,[10] Hsu and Campbell decided that Benjamin was the expert they were looking for to help develop Deep Blue'sopening book, so hired him to assist with the preparations for Deep Blue's matches against Garry Kasparov.[11] In 1995, a Deep Blue prototype played in the eighthWorld Computer Chess Championship, playingWchess to a draw before ultimately losing toFritz in round five, despite playing asWhite.[12]
Today, one of the two racks that made up Deep Blue is held by theNational Museum of American History, having previously been displayed in an exhibit about theInformation Age,[13] while the other rack was acquired by theComputer History Museum in 1997, and is displayed in the Revolution exhibit's "Artificial Intelligence and Robotics" gallery.[14] Several books were written about Deep Blue, among themBehind Deep Blue: Building the Computer that Defeated the World Chess Champion by Deep Blue developer Feng-hsiung Hsu.[15]
Subsequent to its predecessor Deep Thought's 1989 loss toGarry Kasparov, Deep Blue played Kasparov twice more. In the first game of the first match, which took place from 10 to 17 February 1996, Deep Blue became the first machine to wina chess game against a reigning world champion underregular time controls. However, Kasparov won three and drew two of the following five games, beating Deep Blue by 4–2 at the close of the match.[16][17]
Deep Blue's hardware was subsequently upgraded,[3][18][c] doubling its speed before it faced Kasparov again in May 1997, when it won the six-game rematch 3½–2½. Deep Blue won thedeciding game after Kasparov failed to secure his position in the opening, thereby becoming the first computer system to defeat a reigning world champion in a match under standard chess tournament time controls.[20][21] The version of Deep Blue that defeated Kasparov in 1997 typically searched to a depth of six to eight moves, and twenty or more moves in some situations.[22]David Levy andMonty Newborn estimate that each additionalply (half-move) of forward insight increases the playing strength between 50 and 70Elo points.[23]
In the 44th move of the first game of their second match, unknown to Kasparov, abug in Deep Blue's code led it to enter an unintentionalloop, which it exited by taking a randomly selected valid move.[24] Kasparov did not take this possibility into account, and misattributed the seemingly pointless move to "superior intelligence".[21] Subsequently, Kasparov experienced a decline in performance in the following game,[24] though he denies this was due to anxiety in the wake of Deep Blue's inscrutable move.[25]
After his loss, Kasparov said that he sometimes saw unusual creativity in the machine's moves, suggesting that during the second game, human chess players had intervened on behalf of the machine. IBM denied this, saying the only human intervention occurred between games.[26][27] Kasparov demanded a rematch, but IBM had dismantled Deep Blue after its victory and refused the rematch.[28] The rules allowed the developers to modify the program between games, an opportunity they said they used to shore up weaknesses in the computer's play that were revealed during the course of the match. Kasparov requested printouts of the machine's log files, but IBM refused, although the company later published the logs on the Internet.[29]
The 1997 tournament awarded a $700,000 first prize to the Deep Blue team and a $400,000 second prize to Kasparov.Carnegie Mellon University awarded an additional $100,000 to the Deep Blue team, a prize created by computer science professorEdward Fredkin in 1980 for the first computer program to beat a reigning world chess champion.[30]
Kasparov initially called Deep Blue an "alien opponent", but later belittled it, stating that it was "as intelligent as your alarm clock".[31] According toMartin Amis, two grandmasters who played Deep Blue agreed that it was "like a wall coming at you".[32][33] Hsu had the rights to use the Deep Blue design independently of IBM, but also independently declined Kasparov's rematch offer.[34] In 2003, thedocumentary filmGame Over: Kasparov and the Machine investigated Kasparov's claims that IBM had cheated. In the film, some interviewees describe IBM's investment in Deep Blue as an effort to boost its stock value.[35]
Following Deep Blue's victory,AI specialist Omar Syed designed a new game,Arimaa, which was intended to be very simple for humans but very difficult for computers to master;[36][37] however, in 2015, computers proved capable of defeating strong Arimaa players.[38] Since Deep Blue's victory, computer scientists have developed software for other complex board games with competitive communities. The AlphaGo series (AlphaGo,AlphaGo Zero,AlphaZero) defeated topGo players in 2016–2017.[39][40]
Computer scientists such as Deep Blue developer Campbell believed that playing chess was a good measurement for the effectiveness of artificial intelligence, and by beating a world champion chess player, IBM showed that they had made significant progress.[3] Deep Blue is also responsible for the popularity of using games as a display medium for artificial intelligence, as in the cases ofIBM Watson orAlphaGo.[41]
While Deep Blue, with its capability of evaluating 200 million positions per second,[42] was the first computer to face a world chess champion in a formal match,[3] it was a then-state-of-the-artexpert system, relying upon rules and variables defined and fine-tuned by chess masters and computer scientists. In contrast, current chess engines such asLeela Chess Zero typically usereinforcementmachine learning systems that train aneural network to play, developing its own internal logic rather than relying upon rules defined by human experts.[39]
In a November 2006 match between Deep Fritz and world chess championVladimir Kramnik, the program ran on a computer system containing a dual-coreIntel Xeon 5160 CPU, capable of evaluating only 8 million positions per second, but searching to an average depth of 17 to 18plies (half-moves) in themiddlegame thanks toheuristics; it won 4–2.[43][44]
Deep Blue ran under theAIX operating system, and its chess playing program was written inC. Itsevaluation function was initially written in a generalized form, with many to-be-determined parameters (e.g., how important is a safe king position compared to a space advantage in the center, etc.). Values for these parameters were determined by analyzing thousands of master games. The evaluation function was then split into 8,000 parts, many of them designed for special positions. The opening book encapsulated more than 4,000 positions and 700,000grandmaster games, while the endgame database contained many six-piece endgames and all five and fewer piece endgames. An additional database named the "extended book" summarizes entire games played by Grandmasters. The system combines its searching ability of 200 million chess positions per second with summary information in the extended book to select opening moves.[45]
Before the second match, the program's rules were fine-tuned by grandmasterJoel Benjamin. The opening library was provided by grandmastersMiguel Illescas,John Fedorowicz, andNick de Firmian.[46] When Kasparov requested that he be allowed to study other games that Deep Blue had played so as to better understand his opponent, IBM refused, leading Kasparov to study many popular PC chess games to familiarize himself with computer gameplay.[47]
Deep Blue used customVLSI chips toparallelize thealpha–beta search algorithm,[48] an example ofsymbolic AI.[49] The system derived its playing strength mainly frombrute force computing power. It was anIBM RS/6000 SP, asupercomputer with amassively parallel architecture based on 30PowerPC 604e processors and 480 custom600 nmCMOS VLSI "chess chips" designed to execute the chess-playing expert system, as well asFPGAs intended to allow patching of the VLSIs (which ultimately went unused) all housed in two cabinets. The chess chip has four parts: the move generator, the smart-move stack, the evaluation function, and the search control. The move generator is a 8x8combinational logic circuit, a chess board in miniature.[50][51][52][53]
In 1997, Deep Blue was upgraded again to become the 259th most powerfulsupercomputer according to theTOP500 list, achieving 11.38GFLOPS on theparallel high performance LINPACK benchmark. Deeper Blue was capable of evaluating 200 million positions per second, twice as many as the 1996 version.[54]
^Stuart Russell and Peter Norvig, Stuart Russell and Peter Norvig (2020).Artificial Intelligence: A Modern Approach (4th ed.). Pearson. p. 3.ISBN9780134610993.
^LC Catalog – Item Information (Full Record).LCCN2017304768.
^Silver, Albert (19 February 2015)."Deep Blue's cheating move".Chess Base. Chess News.Archived from the original on 29 July 2020. Retrieved3 June 2020.
^Schulz, André (23 November 2006)."Das letzte Match Mensch gegen Maschine?" [The last man vs machine match?].Der Spiegel (in German). Translated by ChessBase Chess News.Archived from the original on 16 October 2012. Retrieved18 August 2021.
Barrat, James (2013).Our Final Invention (Kindle ed.). St. Martin's Press.ISBN978-0-312-62237-4.
Campbell, Murray (1998). "An Enjoyable Game". In Stork, D. G. (ed.).HAL's Legacy: 2001's Computer as Dream and Reality. Cambridge, Mass: MIT Press.ISBN978-0-262-19378-8.
Hsu, Feng-hsiung; Campbell, Murray; Hoane, A. Joseph Jr. (1995)."Deep Blue System Overview"(PDF).Proceedings of the 9th International Conference on Supercomputing. 1995 International Conference on Supercomputing. Association for Computing Machinery. pp. 240–244.doi:10.1145/224538.224567.ISBN978-0-89791-728-5. Archived from the original on 17 October 2018 – via top-5000.nl.
Hsu, Feng-hsiung (2002).Behind Deep Blue: Building the Computer that Defeated the World Chess Champion (1st ed.). Princeton University Press.ISBN978-0-691-09065-8.
.jpg)
