Atlas record: https://www.aitimeline.dev/en/events/evt-0069/

Research · March 1950

Placed by the contemporary primary publication. The exact event date is not known; its documented interval appears below.

Programming a Computer for Playing Chess

Shannon set out how a machine could play chess: an evaluation function for positions, a search some moves ahead, and a minimax choice.

Why it matters

A task treated as proof of thought was reduced to search with evaluation, and that shaped game programs for half a century.

The paper is entirely theoretical; Shannon programmed no machine. He separates a type A strategy, full search to a fixed depth, from type B, which examines only plausible moves, and notes that humans play closer to the second. Both lines were pursued separately afterwards.

Event record

Event date

March 1949 – March 1950

Timeline date

Primary publication date

Verification

Sources gathered automatically · September 17, 2026

Lines

Search and reasoning

ID

evt-0069

Presented at the IRE convention on 9 March 1949 and published in March 1950. The anchor is the publication.

Claude Shannon

Sources

primary Programming a computer for playing chess (publisher record)
Taylor and Francis · Published March 1950
archive Programming a Computer for Playing Chess
Philosophical Magazine · Published March 1950
archive Programming a Computer for Playing Chess (copy of the Philosophical Magazine article)
Computer History Museum · Published March 1950

Related events

Earlier

Builds on Theory of Games and Economic Behavior 1944
Position evaluation and move search rest on the minimax solution from game theory.
- Programming a Computer for Playing Chess · Philosophical Magazine

Later

Related Shannon's relay maze machine 1951links to this record
The same author: Shannon set out a chess program in 1950; the maze is another of his attempts at trial-and-error search.
Related Prinz's chess program 1951links to this record
The program implements part of the scheme Shannon described a year earlier.
Related Digital computers applied to games 1953links to this record
The chapter treats the same task as Shannon's paper, from the side of a paper program.
Builds on Newell's chess machine 1955links to this record
The talk starts from the search and evaluation scheme Shannon described.
- The chess machine: an example of dealing with a complex task by adaptation · Western Joint Computer Conference
Builds on Chess on the IBM 704, full board 1958links to this record
The survey by Newell, Shaw and Simon says the Bernstein program «is in the Shannon tradition»: an evaluation of positions, a search a few moves ahead and a minimax, with the selection of moves as the new step.
- A. Newell, J. C. Shaw, H. A. Simon, «Chess-Playing Programs and the Problem of Complexity», RAND Corporation paper P-1319, 4 September 1958 («To be published in the IBM Journal of Research and Development») · RAND Corporation (copy: bitsavers.org through the Internet Archive)
Builds on A chess program built from goals, not numbers 1958links to this record
The paper names «the basic factorization introduced by Shannon» among the positive features still visible in the new program.
- A. Newell, J. C. Shaw, H. A. Simon, «Chess-Playing Programs and the Problem of Complexity», RAND Corporation paper P-1319, 4 September 1958 («To be published in the IBM Journal of Research and Development») · RAND Corporation (copy: bitsavers.org through the Internet Archive)
Related Mac Hack VI at a tournament 1967links to this record
The program implements Shannon's selective type B strategy.