Discipline and initiative: Turing sounds, here, like a headmaster delivering a speech at the beginning of the term. In the background one hears the Duchess’s lullaby from Lewis Carroll’s Alice in Wonderland. The difference is that the Turing Primer on How to Raise a Computer requires not just beating the computer when it sneezes but providing the pepper to make him sneeze. In the experimental “pleasure-pain systems” that he outlines, for instance, the pepper comes in the form of a random sequence of numbers to which the computer is required to respond in certain ways by a table of behavior. What Turing calls “pleasure” and “pain” are in fact merely instructions to perform arithmetical operations, with the pain stimulus canceling all tentative entries and the pleasure stimulus making them permanent. The problem is that in the absence of any stimuli, “the machine very soon got into a repetitive cycle. This became externally visible through the repetitive B A B A B. . . . By means of a pain stimulus this cycle was broken.” Emotion is thus privileged in the programming process itself, raising the question of whether Turing has imposed arbitrary emotional signs onto a mechanical (and therefore feelingless) process, or whether perhaps human emotion is at heart far more mechanical than we are inclined to admit. Such a strategy fits well with Turing’s larger effort, in the report, to demystify the human body by describing it in the most mechanistic language possible, as in this comparison of the nerve and the electrical circuit: “Certainly the nerve has many advantages. It is extremely compact, does not wear out (probably for hundreds of years if kept in a suitable medium!) and has a very low energy consumption. Against these advantages the electronic circuits have only one counter-attraction, that of speed.”
The extent to which we ascertain that another entity’s behavior shows intelligence, Turing writes at the end of the report,
is determined as much by our own state of mind and training as by the properties of the object under consideration. If we are able to explain and predict its behaviour or if there seems to be little underlying plan, we have little temptation to imagine intelligence. With the same object therefore it is possible that one man would consider it as intelligent and another would not; the second man would have found out the rules of behaviour.
By way of illustrating this point, he proposes an experiment that foreshadows what would later come to be known as the Turing test. Two rather poor chess players—A and C—are placed in separate rooms between which some system of communicating moves has been established. Meanwhile a third man—B—is operating a machine that has been programmed to play chess. C plays a game with either A or the machine operated by B. Will he be able to guess which one is his opponent? Turing suspects that he will find it quite difficult to tell the difference and concludes by remarking parenthetically that the experiment is one he has actually performed. He does not give the result, however, and so “Intelligent Machinery” ends with a number of questions hanging: can a machine, educated through a system of reward and punishment, be said to be able to think? Are children, when they cry or laugh, revealing some spark of soul that distinguishes them from machines, or simply following “rules of behavior” with which we as spectators empathize because we are familiar with them? Or to put it another way, does asking whether computers think require us to ask, as well, whether humans compute?
* * *
*The ACE also used stacks to implement subroutine calls, as do all modern machines. The EDVAC did not. Another innovation not in the EDVAC report is Turing’s “Abbreviated Computer Instructions,” an early form of programming language. I am indebted to Prabhakar Ragde for pointing out these distinctions.
*This was also true of the so-called Johnniacs—the von Neumann–style machines that the EDVAC later inspired.
*In her memoir Mrs. Turing recalled, “The most that Alan told me about his war work was that he had about 100 girls under him. We knew one of these ‘slaves’ as he called them. From her came the information that they marvelled at her temerity in greeting him on Christmas morning with ‘A Happy Christmas, Alan,’ for they held him in great awe, largely because when he rushed into their part of the building on business, he never gave the least indication that he even noticed them. The truth probably is that he was equally alarmed by them.”
*Wilkes was later credited with the invention of microprogramming, for which he received the Turing Award, the Association for Computing Machinery’s highest honor for lifetime achievement.
*The “four-color theorem,” proven in 1997, holds that if you color in a map divided into discrete regions (such as a map of the counties in the state of Florida), you will need a maximum of four colors if no two adjoining regions are to be the same color.
*Turing has slightly altered the story to suit his argument. According to the original account, Gauss’s teacher asked him to add up all the numbers between 1 and 100. His strategy for coming up with the correct answer—5,050—was to divide the hundred numbers in question into the pairs 1 + 100, 2 + 99, 3 + 98, 4 + 97, etc., thus creating fifty pairs, each of which would add up to 101. Gauss then multiplied 50 × 101, obtaining the correct answer.
*That Turing considered his ideas anti-Christian is borne out by the title he gave to a talk he delivered in Manchester in 1951, “Intelligent Machinery, a Heretical Theory.”
7
The Imitation Game
1.
The Manchester in which Alan Turing settled in the fall of 1948 was as noteworthy for its industrial ugliness as for its bad weather. Manchester University, just outside the city center, was equally depressing. In Newman’s laboratory, the walls were covered with brown tiles in what F. C. Williams, his partner in the project, called a “late lavatorial” style. Most of the faculty lived in the suburb of Hale, where Turing rented rooms before buying his first and only house, in 1950, on Adlington Road in Wilmslow, Cheshire. These rooms likely resembled one that W. G. Sebald described in The Emigrants, “carpeted in a large floral pattern, wallpapered with violets, and furnished with a wardrobe, a washstand, and an iron bedstead with a candlewick bedspread.”
The machine on which Turing went to work was a preliminary model intended for small-scale experiments, and thus christened (in keeping with Turing’s educational program) the Baby. It had the distinction, however, of employing Williams’s and Kilburn’s cathode-ray tube technology, which meant that for the first time both the instructions fed into the machine and the results it spit out could be seen. Not that the Baby employed anything so sophisticated as a screen: instead, the numbers appeared in form of bright spots on the monitor tubes themselves. Spots, or “bits,” were arranged on each tube in a 32 × 32 grid (for a total of 1,024 bits), with each bit charged to represent a 0 or 1. A metal pickup plate was set up to detect the charge, and thus “read” the bit’s value. Each 32-bit line in the grid, in turn, represented either a number or an instruction; later, the lines would be lengthened to 40 bits each, with each addressable line containing either one 40-bit number or two 20-bit instructions. As Turing remarked in the programmer’s handbook that he prepared for the Manchester computer, the information in the electronic store could be compared “to a number of sheets of paper exposed to the light on a table, so that any particular word or symbol becomes visible as soon as the eye focusses”—an analogy that recalls the perforated sheets employed at Bletchley in the effort to break the Enigma code.
One of the oddities of working with the Manchester computer was the programming notation with which, as Martin Campbell-Kelly puts it, Turing “saddled users of the machine. . . . Each program instruction consisted of 20 bits, which Turing wrote down as four characters using the 5-bit Post Office teleprinter code. In effect he used the teleprinter code as a base-32 number system. . . .” This in turn required Turing to invent a 32-symbol “alphabet” of number equivalencies in which most numbers were paired with letters—9 was D, for instance; 19 was W—while some were represented by symbols (@ for 2, ″ for 27, £ for 31) and 0 was represented by a slash (/). “Because zero was represented by the forward-strok
e character,” Campbell-Kelly explains, “and this was the most commonly-used character in the written form of programs and data, one early user decided this must be an unconscious reflection of the famously dismal Manchester weather as the effect was that of rain seen through a dirty window pane!” (//////////////) As if things weren’t complicated enough, numbers entered into the machine had to be written backward. Using the base-32 code, the 40-digit binary sequence 10001 11011 10100 01001 10001 11001 01010 10110 (in denary notation, 17 27 5 18 17 19 10 13) would thus have to be written as Z’’SLZWRF—which would, of course, first have to be reversed. This had the effect of leaving anyone who wished to use the machine—including Turing’s assistants, Audrey Bates and Cicely Popplewell—rather beholden to its language teacher. Indeed, when Turing delivered a lecture on “Checking a Large Routine” at Cambridge on June 24, 1949 (the day after his thirty-seventh birthday), his failure to bother to clarify the notational system in which he was writing figures on the blackboard struck Maurice Wilkes, who was in the audience, as “bizarre in the extreme. . . . [Turing] had a very nimble brain himself and so no need to make concessions to those less well-endowed.” The base-32 code was rather like the bicycle that Turing had had at Bletchley, rigged up so that no one but he could ride it.
By way of an experiment to test the efficiency of the Baby, Newman decided to put to it one of the great puzzles of pure mathematics. This involved the so-called Mersenne primes, named after the French monk Marin Mersenne (1588–1648), who in 1644 undertook an investigation into the interesting fact that certain large prime numbers take the form 2n – 1 where n is also prime. As Mersenne soon discovered, the rule did not hold for all prime n’s. (For instance, 2″ – 1 isn’t prime, though 11 is.) However, by the nineteenth century it had been shown that the rule did hold when n was equal to 2, 3, 5, 7, 13, 19, 31, 67, and 127. In 1876 Edouard Lucas (1842–1891) came up with a method by which 2127 – 1 was shown to be prime, and in 1932 D. H. Lehmer (1905–1991) was able to establish that 2257 – 1 was not prime. Subsequently, the Mersenne numbers up to 2521 – 1 were found to be not prime. A number as huge as 2521 – 1, Newman realized, was probably beyond the Baby’s scope; his objective, however, was less to make a discovery than to assess the computer’s capacities. Accordingly, he set the baby to the task of testing Mersenne primes, using Lucas’s method, which required it first to divide the numbers in question into blocks of 40 digits each and then to program the necessary carrying. In the end, though it found no new primes, the Baby was able to verify both Lucas’s and Lehmer’s findings—no mean feat, and a good indication of its potential.*
Operating the Manchester machine wasn’t easy. Among other tasks, the operator had frequently to run from the machine room to the tape room upstairs, where the engineer would, on her instructions, switch the writing current on and then off again. A great amount of physical energy had to be expended, and there was vast room for error. “As every vehicle that drove past was a potential source of spurious digits,” Cicely Popplewell later recalled, “it usually took many attempts to get a tape in—each attempt needing another trip up to the tape room.” Indeed, the members of the Manchester team were soon so lost in the technical complexities of actually getting the machine to do its job that when news of their research reached the press, they were ill-prepared to deal with the consequences. And as it happened, the 1948 publication of a book called Cybernetics, by the American Norbert Wiener (1894–1964), had started a chain of events that cast upon the Manchester project an unwanted spotlight.
The Manchester Computer in 1955. (© Hulton-Deutsch Collection/ CORBIS)
What happened was this: Wiener, who admired Turing, made a special trip to visit him in the spring of 1947 in order to discuss the future of intelligent machines. Wiener’s writings were much more sensationalistic than Turing’s, in addition to which he was something of a futurist manqué, inclined to play up (for instance) the similarity between nerves and electrical circuits and to prophesy scenarios in which robots working at factories render their human counterparts redundant.
Word of Wiener’s ideas and his visit soon reached the ears of Sir Geoffrey Jefferson (1886–1961), the chair of the Department of Neurosurgery at Manchester University and an early advocate of the frontal lobotomy. Jefferson was due to give the Lister Oration at Manchester on June 9, 1949, and chose as his topic “The Mind of Mechanical Man.” In effect, the purpose of the speech was to expose and debunk the Manchester computer project, while hymning the innate superiority of the human soul to anything mechanical or man-made:
Not until a machine can write a sonnet or compose a concerto because of thoughts and emotions felt, and not by the chance fall of symbols, could we agree that machine equals brain—that is, not only write it but know that it had written it. No mechanism could feel (and not merely artificially signal, an easy contrivance) pleasure at its success, grief when its valves fuse, be warmed by flattery, be made miserable by its mistakes, be charmed by sex, be angry or miserable when it cannot get what it wants.
In his report for the NPL, Turing had also addressed, in a rather tongue-in-cheek way, the claim that even if provided with a method of locomotion and sense organs, a machine would still be incapable of enjoying much of what human beings enjoyed. For Turing, however, this was of no consequence: as he later put it, the ability to enjoy strawberries and cream was not a prerequisite for intelligence. Jefferson, on the other hand, brandished the machine’s supposed lack of consciousness as evidence of its ultimate stupidity. Summarizing his oration the next day, the Times of London paraphrased him as saying that unless a machine could “create concepts and find for itself suitable words in which to express them . . . it would be no cleverer than a parrot”; the paper also reported that Jefferson “feared a great many airy theories would arise to tempt them against their better judgment, but he forecast that the day would never dawn when the gracious rooms of the Royal Society would be converted into garages to house the new fellows.”
This was clearly meant as a slight to Newman, whose project the Royal Society had funded, and a day later the newspaper followed up with an article on Newman’s “mechanical brain,” noting that the “mechanical mind” had “just completed, in a matter of weeks, a problem, the nature of which is not disclosed, which was started in the seventeenth century and is only just being calculated by human beings.” The machine was described as being “composed of racks of electrical apparatus consisting of a mass of untidy wires, valves, chassis, and display tubes. When in action, the cathode ray becomes a pattern of dots which shows what information is in the machine. There is a close analogy between its structure and that of the human brain.” The article also included an interview with Turing, who said of the machine,
This is only a foretaste of what is to come, and only the shadow of what is going to be. We have to have some experience with the machine before we really know its capabilities. It may take years before we settle down to the new possibilities, but I do not see why it should not enter any of the fields normally covered by the human intellect and eventually compete on equal terms.
I do not think you can even draw the line about sonnets, though the comparison is perhaps a little bit unfair because a sonnet written by a machine will be better appreciated by another machine!
There was no reason to assume, in other words, that even poetry (Jefferson had ended his oration by quoting from Hamlet) should be the exclusive province of the human imagination. (A relative who read the article told Mrs. Turing, “Isn’t that just like Alan?”) Yet what is more striking than Turing’s willingness to attribute to a machine the capacity for writing and understanding verse is his suggestion that machines might speak between themselves a language no less meaningful for its exclusion of human beings. It was as if what offended Turing, even more than Jefferson’s avidity to shut down avenues of exploration, was his hawking of “humanist” values for the explicit purpose of denying a whole class of beings the right to a mental existence. Likewise homosexual men,
for decades, had been erased from history—and more specifically, from the history of human eros to which Jefferson alluded by mentioning “the charm of sex.” In any case, Turing told the Times, “The university was really interested in the investigation of machines for their own sake.” It was as if, by this point, he was becoming sick of the human.
As for Newman, he gave his own reply to the Times in the form of a letter published on June 14, in which he attempted to summarize some of the science behind the prototype Manchester machine and also clear the air regarding “the rather mysterious description” that the newspaper had given of the problem dating back to the seventeenth century. Testing out the Mersenne primes, he explained, was exactly the sort of pure mathematical exercise at which Newman hoped his machine would excel. Indeed, the earnestness with which he attempted to make the experiment comprehensible to the Times’s readers provided clear evidence as to just how far apart the perspective of the Manchester laboratory was from the one that informed Jefferson’s oration. Nonetheless, the letters column of the Times continued, for a few days, to offer evidence that perhaps Turing and Newman were underestimating the hostility that their research had the potential to provoke. England was as disinclined to accept machine nature as human nature, if Illtyd Trethowan of Downside Abbey, Bath, was to be believed; in a letter to the Times dated June 13, he expressed his hope that “responsible scientists will be quick to dissociate themselves” from Newman’s program. “But we must all take warning from it. Even our dialectical materialists would feel necessitated to guard themselves, like Butler’s Erewhonians, against the possible hostility of the machines.”*
The Man Who Knew Too Much: Alan Turing and the Invention of the Computer (Great Discoveries) Page 20