In 1952, two years after Claude Shannon published his theoretical paper on programming a computer to play chess Alan Turing at Manchester wrote a program for playing chess called the "paper machine," and actually used it in two chess games. At the time there was no machine that could execute the set of instructions that Turing "programmed", so Turing acted as a human CPU, using paper and pencil, requiring more than half an hour to calculate each move. He played a first game against Champernowne's wife, who was a beginner at chess, and won. Then he played one against a colleague, British computer scientist Alick Glennie, which he lost. This second game was recorded.
Turing reported this experiment in his paper entitled "Digital Computers applied to Games" that was published in Bowden's Faster than Thought (1953). Turing's original typescript of this paper was scanned and made available on Google Docs at this link.
See also Reconstructing Turing's "Paper Machine" by Frederick Friedel and Garry Kasparov, available from en.chessbase.com at this link.