Jump to content

Deep Thought (chess computer)

From Alfawiki
Deep Thought
Company typeComputer chess project
IndustryComputer chess, Artificial intelligence
FounderFeng-hsiung Hsu
Headquarters
Pittsburgh
,
United States
Key people
Feng-hsiung Hsu, Thomas Anantharaman, Mike Browne, Murray Campbell, Andreas Nowatzyk
ProductsChess computer systems (Deep Thought, Deep Thought 2)
ParentIBM (after initial development at Carnegie Mellon University)

Deep Thought was a computer created specifically for playing chess, with its initial development taking place at Carnegie Mellon University before continuing under IBM. It represented the second iteration in a lineage of chess machines developed by Feng-hsiung Hsu, beginning with ChipTest and culminating in the creation of Deep Blue. The project team comprised Hsu, Thomas Anantharaman, Mike Browne, Murray Campbell, and Andreas Nowatzyk. Notably, Deep Thought became the first computer to triumph over a grandmaster during an official tournament game by defeating Bent Larsen in 1988; however, it was defeated by Garry Kasparov in a two-game match in 1989 and also lost correspondence games against Michael Valvo.

The system's name was inspired by the fictional supercomputer Deep Thought from Douglas Adams' The Hitchhiker's Guide to the Galaxy, a naming convention that IBM maintained with later programs such as Deep Blue, Deep Fritz, and Deep Junior.

Deep Thought secured victory at the North American Computer Chess Championship in 1988 and claimed the title at the World Computer Chess Championship in 1989, earning a United States Chess Federation (USCF) rating of 2551. Its successor, Deep Thought 2—developed partly by the original team and sponsored by IBM—won the North American Computer Chess Championship once more in 1994, boasting an estimated rating near 2600. The machine employed relatively straightforward evaluation techniques but was capable of examining approximately half a billion positions per move, analyzing up to 10 or 11 moves ahead. Utilizing singular extensions, it could extend analysis along forced move lines even further, once detecting a checkmate sequence spanning 18 moves.

See also

References


External links