336 | 31 COmPUTER CHESS—THE fIRST mOmENTS
‘As I remember, he persuaded me over lunch to take part in his chess experiment’, Glennie
recollected, and so they went upstairs to Turing’s office, ‘a rather bare place with a small table
untidy with paper’. Glennie sat at the chessboard and played his virtual opponent, while Turing
scribbled away calculating the program’s moves, just as he had with Champernowne a few years
previously. He had to keep juggling his various sheets of paper containing the rules: ‘We were
playing on a small table which did not help’, Glennie said. ‘It seemed to go rather slowly and I
think I got slightly bored’.
‘The game took 2 or 3 hours’, Glennie remembered. He won at the twenty-ninth move
(see Game 1). Turing set out the game in his typescript, writing the moves in old-fashioned
chess notation. Beside each move we have placed a translation into modern chess notation (in
square brackets).^21 Turing included some footnotes, and also a column of numbers showing
the increase in position-play value (if any) after each of the program’s moves, and he placed an
asterisk against the number if every other move had a lower position-play value. (In his record
of this game he used O—O to indicate castling, without distinguishing between castling on
the queen’s side and castling on the king’s side; whereas in his Game 2, O—O—O was used to
indicate castling on the queen’s side.) The program, White, opened (see facing page).^22
Talking about the game more than 20 years later, Glennie said ‘I remember it as a rather jolly
afternoon’, adding ‘and I believe Turing must have enjoyed it too—in his way’:
Turing’s reaction to the progress of the play was mixed: exasperation at having to keep to his
rules; difficulty in actually doing so; and interest in the experiment and the disasters into which
White was falling. Of course, he could see them coming.
Glennie explained that in fact Turing had difficulty sticking to the rules:
During the game Turing was working to his rules and was clearly having difficulty playing to
them because they often picked moves which he knew were not the best. He also made a few
mistakes in following his rules which had to be backtracked. This would occur when he was
pondering the validity of White’s last move while I was considering my move. There may have
been other mistakes in following his rules that escaped notice . . . He had a tendency to think he
knew the move the rules would produce.
As will become clear, some of White’s moves in the recorded game were most definitely not the
ones that Turing’s rules dictated.
The mark 2’s mysterious re-game
During his lecture Kasparov introduced a guest, Frederic Friedel, co-founder of ChessBase.
Together, Kasparov and Friedel demonstrated ChessBase’s re-creation of Turochamp. Talking
about Turing’s historic program Friedel said: ‘We have one complete game, played with Glennie’.
‘It’s the only recorded game’, he emphasized. This is not quite true, however.
In 1953, the London publishing company of Pitman and Sons brought out a book with the
intriguing title Faster Than Thought.^23 In effect the world’s first computer science textbook, it
was edited by Vivian Bowden, later Baron Bowden of Chesterfield, who was in charge of com-
puter sales at Ferranti. This book contained contributions from everybody who was anybody
in the emerging world of British computing—including, of course, Turing, who contributed a
section titled ‘Chess’ to the chapter ‘Digital computers applied to games’.^24