# What was Christopher Strachey's draughts program, one of the first game-playing programs?

Date: 2020-01-21 · Author: Robert Scales · https://guidetoartificialintelligence.com/en/learn/what-was-christopher-stracheys-draughts-program/

**Answer:** Christopher Strachey, a schoolteacher at Harrow School in England, wrote a program to play draughts (checkers) against a human, starting on the Pilot ACE computer in 1951 and getting it working on the Ferranti Mark 1 at the University of Manchester by the summer of 1952. It looked a few moves ahead, showed the board on a screen, and is counted among the first game-playing programs ever run on a general-purpose computer.

In 1951 one of the most ambitious programs in Britain was being written in the evenings by a schoolteacher. Christopher Strachey was a schoolmaster at Harrow School, near London, and in his spare time he set out to make a computer play draughts, the game Americans call checkers. By the summer of 1952 his program could play a complete game against a person on the Ferranti Mark 1 at the University of Manchester, displaying the board on a screen as it went.

It is one of the earliest game-playing programs ever run on a general-purpose computer, meaning a machine that can be reprogrammed for any task rather than built for one game. It is also an early example of a computer making its own choices in a contest with a human, which is a small but real step toward artificial intelligence.

## Who Strachey was

Strachey was born in 1916 and studied natural sciences at King's College, Cambridge, where he knew Alan Turing slightly. During the Second World War he did radar research at Standard Telephones and Cables, and afterward became a schoolmaster, first at St. Edmund's School in Canterbury and, from 1949, at Harrow.

His first contact with a stored-program computer came in January 1951. Through a friend they had in common, he met Mike Woodger, who worked at Britain's National Physical Laboratory (NPL) in Teddington. The NPL had just built the Pilot ACE, a smaller working version of a computer design Turing had drawn up after the war.

## From the Pilot ACE to Manchester

Back at Harrow, Strachey decided to write a draughts program for the Pilot ACE. Most beginners would have started with something simple, like a table of numbers. Strachey picked a game, partly inspired by an article on how machines could play chess and noughts and crosses (tic-tac-toe) that NPL scientist Donald Davies had published in *Penguin Science News* in June 1950.

The program was too big. It used up the Pilot ACE's small memory, and when it was first tried at the NPL on 30 July 1951 it failed because of errors in the code.

Meanwhile Strachey learned from Woodger about the Ferranti Mark 1, newly installed at Manchester University, which had a much larger store. Turing, then assistant director of the Manchester computing laboratory, had written its programming manual, and Strachey asked him for a copy. He first visited Manchester in July 1951 and rewrote his program in the Manchester machine's instruction codes, by about October that year.

Turing was impressed but suggested another challenge first: make the computer simulate itself, tracing its own operations step by step. Strachey put the game aside and wrote a trace program of about 1,000 instructions, among the longest written for the machine. He returned to draughts afterward, and by June and July 1952 the program was playing full games at a reasonable speed.

In short, the program was written during 1951, and it first played complete games in the summer of 1952.

## How the program played

The Ferranti Mark 1 stored data on cathode-ray tubes, the same kind of glass screen used in early televisions, and some of those tubes could be watched by the operators. Strachey turned one of these monitor screens into a picture of the draughts board, with the positions of men and kings shown as dots.

To choose a move, the program looked ahead three or four moves, computing the possible positions and picking the one that came out best. Its scoring was simple: count the pieces on each side, with a king worth three ordinary men. The player entered moves with switches and buttons, and the computer printed its own moves on a teleprinter, an electric typewriter connected to the machine.

Strachey also gave it a personality. If the human was slow or made input mistakes, the teleprinter printed messages such as "You must play at once or resign," or, if the human kept ignoring the rules, "I refuse to waste any more time. Go and play with a human being."

Here is the core of that decision process, written as plain steps:

1. List every legal move for the computer.
2. For each move, list every reply, then every reply to that, a few levels deep.
3. At the bottom of each line, score the board: the computer's pieces minus the human's, with kings counting 3.
4. Assume the human always picks the reply that is worst for the computer.
5. Play the move whose worst case is best.

> **Try it:** Set up a draughts board with only four pieces each and play one move using Strachey's scoring (kings count 3). Look two moves ahead for each side and write down the score at the end of each line. Is the move you pick the one you would have played by instinct?

## Why it mattered

Strachey's program was one of the first to show that a general-purpose computer could take part in a game against a person, making decisions and reporting them on a screen and a printer. Some writers call it the first video game for that reason, though the claim depends on definitions, and a noughts-and-crosses program called OXO was running on the EDSAC computer at Cambridge in the same year.

Its biggest influence was on another programmer. In September 1952 Strachey went to an Association for Computing Machinery conference in Toronto and described his draughts program. Arthur Samuel of IBM heard about it there and went on to write his own checkers program for the IBM 701.

The project also launched Strachey's career. Its programming skill impressed the Manchester group, and in June 1952 he left teaching to join the National Research Development Corporation. He became a leading figure in programming language design and helped found denotational semantics, a mathematical way of describing what programs mean.

## How did Strachey's draughts program lead to machine learning?

Strachey's own program did not learn. It played the same way every time, using fixed scoring rules he wrote by hand. He did design a learning routine and discussed with Turing trying it on a simpler game, nim, but he did not complete it before others showed working learning programs.

Arthur Samuel took the next step. Inspired by Strachey's work, he built his own checkers program and kept improving it through the 1950s; in 1956 it was shown on television. Samuel's checkers work became one of the best-known early examples of a computer getting better at a task through experience, the basic idea that machine learning is built on.

That line runs straight to the game-playing systems of recent years, which learn their evaluation of positions from huge numbers of games instead of from a programmer's formula. Strachey's handwritten scoring rule, kings count three, is the kind of human judgment those systems now discover for themselves.

## Sources

- Christopher Strachey: https://history.computer.org/pioneers/strachey.html
- Christopher Strachey: https://en.wikipedia.org/w/index.php?title=Christopher_Strachey&oldid=911382832