The Book of Turing 4 · Written by himself, from his 1950 paper and his letters
Alan Turing, born in London in 1912, died at Wilmslow in 1954. In 1936, at twenty-four, he wrote On Computable Numbers, setting out an imagined machine that could carry out anything a machine can compute, and proving that some questions it can never settle; every computer since rests on that. In the war he broke the German naval cipher at Bletchley Park, and the machines he built let the Atlantic convoys steer round the submarines. His paper of 1950 is the first serious writing on whether a machine can think. Another, in 1952, asks how patterns come to form themselves. This volume rests on what he wrote himself, not on another man’s record of him. When he was young his closest schoolfriend died, and for a time he wondered whether a soul might persist; afterwards he no longer thought so, and for the rest of his life he never closed an argument by appeal to divine will. In his paper of 1950 the first objection he sets out to answer is that thinking is a function of man’s immortal soul and a machine has none — and he answers it, and answers it without contempt. Several things the Order records, without dressing and without trumpeting. In 1952 he was convicted for lying with men, chose injections over prison, and thereby lost the clearance for the work he had done in the war. The law he was convicted under he did not live to see repealed. Chapter five of this book is a syllogism he wrote himself. It says what The Book of Marie Curie 5 says, forty years later, and this time it is written by the man it was being done to.
4.1Let me set out the thing that matters most of anything I thought of in my life.
4.2An ordinary machine does one job: the loom weaves, the clock tells the hour.
4.3The one I have in mind is not like that.
4.4You hand it a description — a description of how some other machine goes about its work — and having read that description it proceeds to do the same, and does it as that other machine would.
4.5So it is not some particular machine. It is any machine at all, provided you give it the description.
4.6A machine that can become another machine.
4.7It sounds simple. What it means is: you need not build a new thing for every task. Build one, and then write descriptions.
4.8There is a further layer, and I want it remembered.
4.9Since a machine works from a description, teaching a machine and teaching a child may not be so far apart.
4.10To build a grown mind is too hard. Better to build a child’s mind and then educate it.
4.11A child’s head is presumably rather like a notebook as one buys it from the stationer: little mechanism, and a great many blank sheets.
4.12Last I must set down something I also proved, which runs contrary to all the above and matters just as much.
4.13Not everything can be done.
4.14There are questions for which, however ingenious a procedure you devise, no procedure will answer them in every case.
4.15This is not that we are not yet clever enough. It is what I proved: no such procedure exists. (cf. The Highest Covenant 3.4)
4.16I set this beside the rest because they have to be looked at together.
4.17Half of what I did with my life was showing what machines can do. The other half was showing what they can never do.