Prof

by Dermot Turing

  Witness: I don’t think you’re serious. By a winter’s day one means a typical winter’s day, rather than a special one like Christmas.

  And so on. What would Professor Jefferson say if the sonnet-writing machine was able to answer like this in the viva voce?

  The debate with Jefferson was not yet finished, however. Alan gave a talk on the BBC’s Third Programme (the predecessor of Radio 3) on the subject in May 1951, and in January 1952 the Third Programme hosted a debate between Turing and Jefferson, with Newman participating as well. The whole thing was introduced by Professor Richard Braithwaite of King’s College, Cambridge – who had, back in 1933, introduced Alan to the Moral Sciences Club and the subject of philosophy. It is still fresh and lively even after more than 60 years; they discuss whether computers can have appetites, complain about the programs they are given, have a sense of duty, or throw tantrums; and it concludes with Jefferson saying, ‘it would be fun some day, Turing, to listen to a discussion, say on the Fourth Programme, between two machines on why human beings think that they think’.

  Meanwhile, in Manchester, a new computer had come into operation. This one wasn’t a baby, it wasn’t at all lousy, and it was too busy to pay attention to thoughtless chatter on the wireless. Too busy writing love letters.

  Notes

  1 Mike Woodger, also hired by the NPL to assist Turing

  2 Senior Principal Scientific Officer

  9

  TAKING SHAPE

  IT WAS ALL STRACHEY’S FAULT. Not Lytton, arguably the most overt of the Cambridge Hellenistic homosexuals, or even his brother Oliver, who was more closely connected with Alan Turing as a veteran of Room 40 and a senior codebreaker at Bletchley Park. This Strachey was Christopher, Oliver’s son, and another former student of mathematics from King’s College, Cambridge. Christopher Strachey had gone up to King’s in 1935, while Alan was doing his work on Computable Numbers, and since the war he had been teaching maths and physics, most recently at Harrow. But in 1951 he was introduced to Alan’s former assistant at NPL, Mike Woodger, and Strachey started producing computer programs. He wasn’t interested in solutions of 300-year-old problems relating to prime numbers. Like Alan, Strachey saw the potential for the computer, and his particular interest was games. But to play even children’s games the computer needed a grown-up memory.

  Since October 1948, when Alan was installed as Deputy Director of the Computing Machine Laboratory, a lot of engineering had been going on. The Manchester baby was being rebuilt as a grown-up computer, to be called the Manchester Mark 1. This was an altogether more professional machine, and the Royal Society grant was put towards a modern building with acceptably low levels of radioactivity in which it could be housed. Technical experts from the Telecommunications Research Establishment – which was not only the home of radar but had an honourable record in producing computing equipment – and a local Manchester firm called Ferranti joined forces to develop a full-capacity machine which Ferranti would then be able to exploit commercially. Unlike the baby, its innards were neatly arrayed and hidden within streamlined metal cabinets. The Mark I looked professional. On the other hand, its programming manual was written by Alan Turing.

  Programming the Manchester Mark 1 was not for the faint-hearted. R.K. Livesey, who was assisted by Alan in 1953 with the mathematical aspects of an engineering problem, recalled:

  Programming the Mark 1 was certainly ‘machine code programming’. Each machine instruction consisted of 20 binary digits. An instruction was fed into the machine as four rows of holes/blanks on 5-track teleprinter tape, each row corresponding to a character on a teleprinter keyboard. Thus the written form of a program consisted of a sequence of ‘words’, each of four teleprinter characters. Unfortunately the characters corresponding to the 32 binary numbers 00000 … 11111 were arranged in the entirely arbitrary sequence /E@A:SIU1/2DRJNFCKTZ

  LWHYPQOBG2MXV£. Anyone who used the machine regularly ended up knowing this sequence by heart – I can still repeat it from memory. And this bizarre programming code was not the only complication facing a user. The Mark 1 had a C.R.T.1 store, and in a C.R.T. store the trace always goes from left to right. So all numbers were stored and processed with the most [sic] significant digit on the left. (I imagine Turing could have changed this in the original specification of the machine, but he probably had no difficulty in doing arithmetic backwards himself and couldn’t imagine it being a problem for anyone else.)

  No surprise, then, that the programming manual was later redone in a more user-friendly way (for example, allowing decimal input) with the help of an assistant called Cicely Popplewell. In February 1951, the new Manchester Mark 1 was switched on, and Alan Turing made his first entry in the machine’s logbook. One innovation of the new computer was a random-number generator: not the kind of thing you needed to churn out trajectories for missile development, but just perfect for computer games. In May, Alan gave his broadcast on the Third Programme. Christopher Strachey was listening in, and fired off a four-page letter about teaching (something about which he had some experience) as applied to machines (about which he was learning); oh, and mentioning his program for the Pilot ACE to play draughts. This was doomed – it had exhausted the small machine’s memory – but the Manchester Mark 1 could do what the Pilot ACE could not. Alan provided the programming manual for the new Manchester Mark 1 computer, and Strachey translated his program. He also persuaded the computer to play Baa, Baa, Black Sheep. The random-number generator could also be deployed in the serious business of love letters:

  At the console. Alan Turing leans over the console of the rather swanky new Mark 1 computer in Manchester.

  Darling Sweetheart

  You are my avid fellow feeling. My affection curiously clings to your passionate wish. My liking yearns for your heart. You are my wistful sympathy: my tender liking.

  Yours beautifully

  M. U. C.1

  Honey Dear

  My sympathetic affection beautifully attracts your affectionate enthusiasm. You are my loving adoration: my breathless adoration. My fellow feeling breathlessly hopes for your dear eagerness. My lovesick adoration cherishes your avid ardour.

  Yours wistfully

  M. U. C.

  All this was a step on the road to the programmer’s Holy Grail: to write a program which could play a decent game of chess. Ever since those discussions in the pub in Wolverton had this been an objective. Donald Michie had been trying to write a chess algorithm:

  Alan told me that he and Champernowne had constructed a machine to play chess, in the sense of a complete specification on paper for such a machine. One could call it a ‘paper machine’ from which one could laboriously calculate move by move what the corresponding electronic machine would do were it constructed. Each move required perhaps half an hour’s paper work as compared with the fraction of a second which a real machine would need. During a stay in Cambridge, Shaun Wylie1 and I constructed a rival paper machine which we christened ‘Machiavelli’, from our two names, Michie-Wylie. On behalf of Machiavelli we then issued a challenge to the Turochamp (our name for the Turing-Champernowne machine), the game to be played by correspondence.

  I.J. Good had found out about the Machiavelli already.

  16 Sep 48

  Dear Prof,

  Pardon the use of the typewriter: I have come to prefer discrete machines to continuous ones.

  I visited Oxford last week-end. Donald showed me a ‘chess machine’ invented by Shaun and himself. It suffers from the very serious disadvantage that it does not analyse more than one move ahead. I am convinced that such a machine would play a very poor game. […]

  Yours, with best wishes,

  Jack

  Sept 18, 1948

  Dear Jack,

  The chess machine designed by Champ & myself is rather on your lines. Unfortunately we made no definite record of what it was, but I am going to write one down definitely in the next few days with a view to playing the Shaun-Michie machine. […] />
  Yours

  Prof

  History doesn’t reveal whether the TuroChamp bested the Machiavelli. It does reveal that Champ could find Alan exasperating at times:

  My wife and I had invited Alan to stay with us at Shotover around Christmas. One morning an envelope arrived containing a piece of perforated tape, and the postmark (Manchester) led me to suppose that this was Alan’s eventual response to our invitation. Four hours of hard work broke the code and I learnt that he would arrive at 2 a.m. the next morning, and that a parcel of food which he was sending must be unpacked and immediately dealt with according to some specified instructions. My satisfaction in deciphering the message was damped the next day when Alan explained it had only taken him half a minute to type the message on to the tape, as it was in standard teleprinter code, and I gathered he had hardly supposed it would occupy more than a few minutes of my time to reverse the process.

  With the power of the Manchester Mark 1, moving from a paper algorithm to an electronic program became a possibility. The time was also right and everyone was onto the problem: Claude Shannon had written his paper in 1950, as had one of the computer experts from the NPL (Donald Davies, who had been able to deliver the working Pilot ACE). Alan began work on converting the TuroChamp into code, but the task wasn’t completed. Instead, Dietrich Prinz, one of the Ferranti engineers, wrote a working chess program which ran on the Manchester machine in 1951. Alan’s own written contribution to the chess-program literature came as part of a chapter on ‘Digital Computers Applied to Games’ published in 1953.

  Chess Champ. David Champernowne in about 1959: Alan Turing’s lifelong friend, mathematician, economist, and co-designer of the chess-playing algorithm TuroChamp.

  Jack Good’s correspondence with Prof wasn’t just about computers playing games. For example:

  3 Oct 48

  Dear Prof,

  I have just read Adrian’s ‘The physical background of perception’ (Oxford, 1947, pp.96: lectures at Magdalen, Oxford). Here it is stated that there are 10ooo000ooo nerve-cells in the integrated nervous system. Presumably the vast majority are in the brain. There is an interesting passage here: ‘… we can still accept the hypothesis that the physical basis of a memory is in the nature of a resonance pattern which may be established in local circuits throughout the whole of the cortex’. I like the idea of resonance patterns in spite of its vagueness.

  Have you heard of the TRANSISTOR (or Transitor)? It is a small crystal alleged to perform ‘nearly all the functions of a vacuum tube’. It might easily be the biggest thing since the war.

  With best wishes and good luck in your new job,

  Jack

  While all the fuss was going on about thinking machines, considerable thought was going into questions of neurology. It was no accident that the wireless debates had involved Sir Geoffrey Jefferson: certainly, he had weighed in on the question of sonnets, but as one of the country’s foremost brain surgeons who had patched up soldiers with head wounds in two world wars, he had more claim than most to know what he was talking about. The other side of the question was how the brain thinks, and how it controls the body. If you regarded the body as a kind of machine …

  Pattern recognition

  In 1948 Norbert Wiener published a book. Wiener was a mathematician at the Massachusetts Institute of Technology – with the impeccable academic lineage of having studied with Russell in Cambridge and Hilbert in Göttingen; later on he worked with Claude Shannon, John von Neumann and others connected with the early development of computers. Wiener’s book was called Cybernetics, or control and communication in the animal and the machine. Wiener opens up the crossover area where machines and animals have similar characteristics. He discusses how animals perceive and recognise things (by sampling, the same way that Shannon’s proposal for voice encryption worked, perhaps?); he explores methods of communication and language, and he compares computers and the nervous system. He also uses plenty of equations.

  We have already spoken of the computing machine, and consequently the brain, as a logical machine. It is by no means trivial to consider the light cast on logic by such machines, both natural and artificial. Here the chief work is that of Turing. [Wiener cites Computable Numbers.] We have said before that the machina ratiocinatrix is nothing but the calculus ratiocinator of Leibniz with an engine in it; and just as modern mathematical logic begins with this calculus, so it is inevitable that its present engineering development should cast a new light on logic.

  Wiener had visited NPL and discussed the idea of a thinking machine, or machina ratiocinatrix, with Alan Turing then, even though there was no hardware realisation of it at that stage. Despite the shortage of hardware, and the abundance of Latin, the ideas in Norbert Wiener’s book had caught on. On a stifling September evening in 1949, a group of neuroscientists, engineers and physicists were concluding the inaugural meeting of a new society, called the ‘Ratio Club’, in an echo of Wiener’s Latin tag. The purpose of the society was to discuss the new topic of cybernetics. It had few rules: one was that there should be complete freedom of expression, and another was that there should be no professors (because professors induce deference). It was invitation-only, and slightly antiestablishment; perhaps a bit like the Apostles, except without the aesthetes, spies and pretentiousness. As the first meeting broke up, it was suggested that some mathematicians be invited to join their number, to ‘keep the biologists in order’. Alan Turing (who was still not a professor) was suggested, the proposal was unanimously supported, and thereafter Alan assiduously attended its meetings. Later in 1950 I.J. Good also joined. In April and December 1950 Alan gave talks. One was on ‘Educating a Digital Computer’, covering the issues discussed in Computing Machinery and Intelligence, and was ‘remembered as being particularly good with Turing on top form stimulating a scintillating extended discussion’.

  Yet computing and machine intelligence were only part of what was interesting at the Ratio Club meetings. Some of the other presentations given to the Club heralded a shift in the direction of Alan’s own interests. The biologists were giving new order to the mathematicians.

  • 19 January 1950: ‘Why is the Visual World Stable?’ The presenter was Donald Mackay, a physicist interested in both machine intelligence and neuropsychology.

  • 16 March 1950: Introductory talks from Ross Ashby and Horace Barlow. Barlow was a neuroscientist, a nephew of Sir Charles Darwin, but a member of the Club for his expertise in vision and neurology. Ashby had been corresponding with Alan Turing in 1946, about the potential for modelling adaptive processes on the unbuilt ACE, and writing papers, of which Alan Turing had a good collection, on subjects like adaptation and neural networks. One was entitled Design for a brain. Another in the collection is Ashby’s later paper from 1952, entitled Can a mechanical chess-player outplay its designer?

  The Ratio Club. This high-energy group of non-professors debated the crossover area between maths, computing and biology. Back row (l-r): Harold Shipton, John Bates, W.E. Hick, John Pringle, Donald Sholl, John Westcott, Donald Mackay. Middle row: Giles Brindley, Tom McLardy, Ross Ashby, Thomas Gold, Albert Uttley. Front row: Alan Turing, Gurney Sutton, William Rushton, George Dawson, Horace Barlow.

  • 18 May 1950: ‘Pattern Recognition’. The presenters included Grey Walter, a natural sciences graduate of King’s College, Cambridge, who had overlapped with Alan as a student there, and was now working on his invention, a robotic ‘tortoise’ which could find its way back to its ‘hutch’ where it could recharge its electric supply; Albert Uttley, who had worked at the Telecommunications Research Establishment during the war, developing computing equipment; also Donald Mackay, Thomas Gold (then developing a theory on the workings of the inner ear) and Horace Barlow.

  • 21 September 1950: ‘Noise in the Nervous System’. The presenter was John Pringle, who had been an undergraduate and fellow alongside Alan at King’s. He had proposed Alan for membership of the Club, and was now a neurobiologist.

 
• 5 April 1951: ‘Shape and Size of Nerve Fibres’. The presenter was William Rushton. His work was on electrical excitation of nerve cells, but he was working on vision and, in particular, colour-blindness.

  By the beginning of 1951 Alan’s interest in patterns in biology had surfaced in a correspondence he was conducting with a professor, and who was therefore barred from the Ratio Club. This was the scientist Professor J.Z. Young, best remembered for his experiments with the giant nerve of the squid. Young had given the 1950 Reith Lectures on the BBC, on the subject of ‘Doubt and Certainty in Science’. Lecture 2 was called ‘Brains as Machines’; Lecture 7 was entitled ‘The Mechanistic Interpretation of Life’. Among the influential sources cited by Young were someone called ‘A.S. Turing’ and Norbert Wiener. Part of the correspondence between Young and Turing was about the ability to recognise things:

  Dear Turing,

  I have been thinking more about your abstractions & hope that I grasp what you want of them. Although I know so little about it I should not despair of the matching process doing the trick. You have certainly missed a point if you suppose that to name a bus it must first be matched with everything from tea-pots to clouds. The brain surely has ways of shortening this process by the process – I take it – you call abstracting.

  Yours

  John Young

  Dear Young,

  I think very likely our disagreements are mainly about the uses of words. I was of course fully aware that the brain would not have to do comparisons of an object under examination with everything from teapots to clouds, and that the identification would be broken up into stages, but if this method is carried very far I should not be inclined to describe the resulting process as one of ‘matching’. […]