COPElAND & PRINz | 331
quite easy to find instruction tables [programs] which would enable the ACE to win against an
average player’.^11
Then, in the summer of 1948, Turing described a chess-player version of what we now call
the Turing test or imitation game (see Chapters 25 and 27). He wrote:^12
It is not difficult to devise a paper machine which will play a not very bad game of chess. Now get
three men as subjects for the experiment A, B, C. A and C are to be rather poor chess players,
B is the operator who works the paper machine. (In order that he should be able to work it fairly
fast it is advisable that he be both mathematician and chess player.) Two rooms are used with
some arrangement for communicating moves, and a game is played between C and either A or
the paper machine. C may find it quite difficult to tell which he is playing. (This is a rather ideal-
ized form of an experiment I have actually done.)
The ‘paper machine’ that Turing mentioned was created that same summer in collaboration
with his friend the statistician David Champernowne. Turing crystallized his ideas about com-
putational chess into a kind of program in pseudo-code—a set of machine-rules for playing
chess. Named ‘Turochamp’, and an outgrowth of his wartime deliberations with Michie and
Good, this was an early manifestation of what would now be called an AI program.
Turochamp
Turochamp was a ‘paper machine’ in the sense that since no electronic computer was available,
Turing and Champernowne simulated the program’s play by hand, using paper and pencil. In
1980 Champernowne gave this description of Turochamp (quoted in The Essential Turing):^13
It was in the late summer of 1948 that Turing and I did try out a loose system of rules for decid-
ing on the next move in a chess game which we thought could be fairly easily programmed for a
computer . . . Here is what I think I remember about the system but I may have been influenced
by what I have since read about other people’s systems. There was a system for estimating the
effects of any move on White’s estimated net advantage over Black. This allowed for:
(1) Captures, using the conventional scale of 10 for pawn, 30 for knight or bishop, 50 for rook,
100 for queen and something huge, say 5000, for king.
(2) Change in mobility; i.e., change in the number of squares to which any piece or pawn could
immediately move legitimately (1 each).
(3) Special incentives for: (a) Castling (3 points). (b) Advancing a passed pawn (1 or 2 points).
(c) Getting a rook onto the seventh rank (4 points perhaps).
(I don’t think occupation of one of the 4 central squares gained any special bonus. We did not
cater to the end-game.) Most of our attention went to deciding which moves were to be fol-
lowed up. My memory about this is infuriatingly weak. Captures had to be followed up at least
to the point where no further capture was immediately possible. Checks and forcing moves
had to be followed further. We were particularly keen on the idea that whereas certain moves
would be scorned as pointless and pursued no further others would be followed quite a long
way down certain paths. In the actual experiment I suspect we were a bit slapdash about all this
and must have made a number of slips since the arithmetic was extremely tedious with pencil
and paper. Our general conclusion was that a computer should be fairly easy to programme to