LLMpediaThe first transparent, open encyclopedia generated by LLMs

Deep Blue

Generated by DeepSeek V3.2
Note: This article was automatically generated by a large language model (LLM) from purely parametric knowledge (no retrieval). It may contain inaccuracies or hallucinations. This encyclopedia is part of a research project currently under review.
Article Genealogy
Parent: The Intentional Stance Hop 4
Expansion Funnel Raw 51 → Dedup 0 → NER 0 → Enqueued 0
1. Extracted51
2. After dedup0 (None)
3. After NER0 ()
4. Enqueued0 ()
Deep Blue
NameDeep Blue
ManufacturerIBM
TypeSupercomputer
Release date1996
PredecessorDeep Thought

Deep Blue. It was a supercomputer designed by IBM specifically to play the game of chess. Its historic victory over world champion Garry Kasparov in 1997 marked a pivotal moment in the history of artificial intelligence and human–computer interaction. The project was a direct successor to the earlier Carnegie Mellon University system known as Deep Thought.

Overview

The primary objective of the IBM Research project was to build a machine capable of defeating the reigning World Chess Champion under standard FIDE tournament conditions. The team, led by developers such as Feng-hsiung Hsu and Murray Campbell, combined specialized VLSI chess chips with a massively parallel processing architecture. This effort represented a significant milestone in the field of computer chess, shifting research from purely software-based approaches to custom hardware solutions. The matches against Garry Kasparov captured global media attention, turning the events into a symbolic battle between human intellect and machine calculation.

Development and design

The project originated at Carnegie Mellon University with the creation of Deep Thought in the late 1980s. Key researchers, including Feng-hsiung Hsu, later moved to IBM Research to continue development. The final system, housed at the Thomas J. Watson Research Center, utilized a RS/6000 SP2 computer as its host, coordinating 480 custom-designed VLSI chips. These chips functioned as chess engine accelerators, capable of evaluating up to 200 million positions per second. The software incorporated an opening book compiled by grandmaster Joel Benjamin and complex evaluation functions that could assess positional nuances beyond simple material count.

Matches against Garry Kasparov

The first six-game match under match conditions occurred in Philadelphia in 1996, where Garry Kasparov won 4–2, despite losing the first game. The rematch was held in New York City in May 1997 under the auspices of the Association for Computing Machinery. After a dramatic series of games, including a famous victory for Kasparov in the first game and a critical blunder in the final game, Deep Blue won the match 3.5–2.5. The final game, a Sicilian Defence, ended with Kasparov resigning after just 19 moves. The outcome sent shockwaves through the chess world and sparked widespread debate in publications like The New York Times and Scientific American.

Impact and legacy

The victory is widely considered a landmark event in the history of artificial intelligence, demonstrating that machines could outperform humans in complex intellectual tasks with defined rules. It influenced subsequent developments in AI, including projects like Watson and AlphaGo. Within chess, it accelerated the adoption of computers as essential tools for preparation and analysis by top players like Vladimir Kramnik and Viswanathan Anand. The original machine was retired after the 1997 match, and parts of it are housed at institutions like the National Museum of American History and the Computer History Museum.

Technical specifications

The system was a hybrid MIMD architecture combining general-purpose IBM POWER2 processors with application-specific integrated circuits. It featured 30 IBM RS/6000 nodes, each containing multiple VLSI chess processors, for a total of 480 dedicated chips. Its software ran on the AIX operating system and could search an average of 100 to 200 million positions per second, reaching depths of up to 12 plies in complex middlegame positions. The evaluation function included over 8000 features, weighing factors from pawn structure to king safety, guided by the advisory work of grandmaster Joel Benjamin.

Category:Supercomputers Category:Chess computers Category:IBM