Alan Turing: an Introductory Biography

Transcription

1 Alan Turing: an Introductory Biography Andrew Hodges Wadharn College, University of Oxford Summary. A short d( 'script.ion of the events a nd issues in t he life of Alan Turing ( ). The Turing Day collference at the Swiss Federal hlstitute of Technology, Lausanne, was held to mark the nilletieth a.jini ve ninry of A lan T millg's birt h, which fell on 23 rd June Turing's life was so s llort thai ful'ther events will soon ma.rk the fift ieth anniv('l'sary of his death Oll 7th.June But in that span between 1912 and 1954 Alall Turiug did pionc('ring IVOl"k, ellcompassing the founda.tions of computer science, wliich still contillues to stitllulate and in spire. As this volume illustrates, the breadth and depth of Turing's wol'k, as well as itci dramatic int.ensity, colnpellsates for its chrollological brevil y. Ala n Turing's biography is interwovell witll the COUUiC' of twent.icthcentury hist.ory nne! falls naturally iuto pre-war, wartime and post-war periods. He was born iuto the Drihih Ilppel'-tlliddlc c1acis wllicit ha d confidently r un the iillperial administratiotl unt.il the Fil'st. \ivorld vvar, but which, under the impact of ecollomic a.nd political nisi::;, progr e~s ivel y lost control thereafte r. In a v('ry broad sellsc'. Alall T lll'ing helollged to a new) modernizing generation II'!tieh remted COlltClllpl'ltOusl y agaillst Victorian va.llles. Dut Alan Tmillg's cady life was llliuked by detaclimcnt from t.he obligatory social trailling, rather thall rclw ilioll ag(lillst it. It was also marked from the start by his iutcnsely individua l respollse to science and mathematics, in particular to the relativity a1ld quhntulll mechallics which h ad transformed t.he physical sciclices since He bcc:alllc a ll undergradua te at King's College, Cambridge Ullivcrsity, ill 1931, reading mathem atics and graduating wi th dist.inctioll in Very soon, in 1935, the lcctme:o of M. H. A. (ji, Iax) Newlila n at Calllbridge introduced him to the frontier of mat.hematical logic, Ivhic!t likewise had been transformed s ince But logic WHS llei ther T1U'illg '::; illll1lecl iat.e nor his only choice. It was his work in proba bilit.y theory that \NOIl him a Fellowship of King's College ill 1935, anc! he migltl' easily have coiltinued in this field, or in the m athematical physics that bad first. at tracted him. Thu:o he came to logic: from a wide background in P Ul'C a ue! a pplicd ma thema.tics, a ud it was in this eclectic spirit that he auacked the E ntscheidungsproblem of David Hilbert, which a t t.ha.t point rel11a.illed all outstanding quest. ioll.

2 4 Andrew Hodges Turing, working alone, and only twenty-three, attacked and settled this problem using his definition of computability. His fa.mous paper, "On Computable Numbers, with an Applica.tion to the Entscheidllngsproblem," was published at the turn of A complete outsider to the fi eld. he won a place in the subject "lith a collcept which after 60 years remains definitive. His defini tion of computability showcd there could be no ge!leral method for deciding the provability of mathematical proposit.ions, and ma rked the ('!lei of at.tempts to forma.lize a. complete systelll for mathematics. But it also opened the way into new fi elds, which now wc would recognize as computer science and the cogniti ve sciences. A lthough Turing thereafter found himself classed as a logician, he was more a mathematicia.n who applicd himself to logic: and mole than that, a scientist who behind the mat.hemat.ics felt. a dcep concel'll for the fundamental questions of mind and mattcr. His underlying interest ill the problcm of mind showed up in the bold st.atements about humall lilell10ry and st.atcs of m ind which informed his argluueilts. His backgrounci in physics was hillted ect in the "machincs" with which hc lilade his clefinition of computability the now-famous "Turing machil\es," l'1lnning 011 paper tape, an inwge of 1930s modernity. It was this concreten('ss which mae!e Turillg's defi nition of COlllputability much more satisfactory t. ha.ll the lllclthell1atical dcfillit ioll offered hy Alonzo Church, the Princeton logician wlio lee! the fielcl. Matlwlllatically, Turing's defini tion WeiS equiva.lellt to Church':,;. But the descriptioll of the Turing machine gave a convincing arglllliell t. for why it \ovas that this mat.hclilatica.l ctcfinition complet.ely captured the concept of "cffect.ively calculable." Each Turillg machine rcprescuts all algorithm; fo r 1ll0derll reaclcrs it. is hard not to see it as a computcr program ane! t.o bc(\!' ill mind that COIll )Juters did not t hen exist. But T urillg specifica.!ly dcfillcd it type of l1lacllillc call ed "universal," capable of reae! ing t.he illstl'llctioll tablc of ally ot.hel llw, chine. This is precisely the principle of the stored-progralll digital colllpu ter, then yet to corne into being. It is possible t hat. Tmillg evell then clltcl't.aillecl the possibility of constructill g s Llch a. nw.chillc, for he certaillly illterested himself in electrical a nd mecllallical COlllPutation. Bllt, if so, he left 110 not.cs 01' observations OIl this questioll. Rcither. be was prilllarily engagcd ill i\ wide variety of mathematical researches. In lat.e 1036 Tlll'ing joillcd Cltlll'ch \ group at Princeton and there embarked on mole' advanced logic bllt a lso Oil work in a lgebra and on developing thc theory of the Ricmallu zeta-function, flllldamental to the study of primo llumbers. Tlte ll1atij('llla.tician John von Nellmann offerecl him a post at Princet.on t.o continue lll<'lthellla.tical research, bnt he chose to return to England in slimmer 1938, consciolls of the impendiug conflict with Germany and already prepared to ma kc a special contribut.ion to it. Whilst the Seconcl vvorlci vvar took ma ny of ltis scient ific contemporaries into the physics of radar a nd the atomic bomb, it. took Alan Turing into n yptology. Aft.er 1938, his grappling with the illfini tudes of mathcnl<1l'ical logic

3 Introductory Biography 5 was complemented by the finite but still highly challenging logical problem of the German Enigma enciphering machine. In 1939, partly thanks to a brilliant Polish contribution, Turing was able to propose a highly ingenious method of testing a "probable word" for Enigma-enciphered messages. His logical scheme was rapidly materialized in very large electromechanical devices called Bombes, which from 1940 onwards vvorked as the central engines of decipherment throughout the war. For this work, Turing was based at the now famous center at Bletchley Park, Buckinghamshire, which recruited increasingly large sectors of the British intelligentsia. Amongst these, Alan Turing remained the chief scientific figure. His central contribution, after the logic of the Bombe, lay in Bayesian statistics for measuring "weight of evidence," a development close to Shannon's theory of information measure. Turing led what was in effect a scientific revolution, and because he took personal charge of the crucial U-boat message problem, was able to see his approach triumph in the battle of the Atlantic, Alan Turing's role mirrored the developing course of the war: at first a lone British figure against all the odds, and later, as the work developed on a major industrial and transnat.ional scale, handing over tl1e British colltribution to the power by which it was eclipsed: the United States. Turing'S personality traits became more striking when outside the Cambridge environment; shy but outspoken, nervous but lacking deference, he was liot well adapted to military manners or to the diplomacy of the embryonic Anglo-American relationship. But his commanding scientific authority made him the top-level technical liaison between the wartime Allies, demanding a voyage to America in the winter of at the height of the Atlantic batlie. None of this experience, however, gave him a taste for power or detracted from his primary vocation as a pure scientist. The undiminished tenacity of his scientific calling was well illustrated by the lise he made of his wartime experience. For after 1943 Thring knew from Bletchley Park work that largescale digital electronic machinery had the speed and reliability to make possible a practical version of his "uni versal machine." F\'01l1 that point onwards he made the construction of such a machine his prillcipa.l ambition, and he arranged his work so as to gain persona.! experiell(,(' of electronic components - designing and building an advanced speech scrambler. And so, at the end of the Second \ii/orld War, he had a plan for an electronic computer, but it was motivated not by military or economic needs. It was for the exploration of the scope of the computable and in particular for comparing machine processes with huma n mental processes. He called it "building a brain." For his war work, which some would judge critical t.o the Atlantic war, Thring was honored with the modest British formality of an OBE. But his work remained completely secret until the mid-1970s, and he derived no advantage from it in his subsequent scientific career. Nevertheless, the post-war period began with great promise, for he was invited to take up an appointment at the National Physical Laboratory, near London, in October 1945,

4 6 Andrew Hodges and his electronic computer plan, the proposal for the Automatic Computing Engine (ACE), was swiftly adopted in IvIarch At that time, which was before the word "computer" had its modern meaning, Turing l -;ed the term Practical Universal Computing Ivlachine. But, although fond of the word "practical," Turing did not have the human gift of getting his practical way with people and institutions who did not share his vision. From t he outset, it became clear that the NPL had no clear idea on how it was to build the machine he had designed, and it failed to adopt a policy speedy enough to satisfy Alan Turing. Turing's plans for software, exploiting the universality of the machine, were t he strongest feature of his proposaj, but they were little developed or publicized because of the dominat.ing problem of hardware engineering. Impatient fol' progress, Turing took up marathon running to near-olympic standard, but this did not relieve the stress. In t.he autumn of 1947 he returned to Cambridge for a sabbat.ical year, and while there was approached by Max Newma n, since 1945 professor of rnathemat. ics at Manchester Uniwrsity, to take an appointment there instead. Newman had played a most important part at Bletchley Park after 1942 and had organized a section using the most sophisticated elect.ronic machinery; he was also fu lly acquainted with Turing's logical ideas. At Manchest.er he had rapidly recruited both Royal Society funding and top-rank engineers, and by June 1948 a tiny version of the universal machine principle was working there in marked contrast to the lack of progress at the NPL. Turing accepted the appointment as Deputy Director of the Computing Laborat.ory. But already in 1948 it became clear that t.he engineering wo uld dominate the Manchest.er environment, and before long both,\lewman and Turing were sidelined and did not direct anything at all. Turing's programming never exploited the advanced possibilities he had mapped out in 1946, and he failed also to write the papers that. could have established his claim to the theory and practice of modern comput.ing. Instead, the main theme of his work became the more futurist.ic prospect. of Artificial Intelligence, or "intelligent machinery" as he called it. Already prefigured in 1946, this was expounded in papers of 1947, 1948, and 1950, arguing strongly that computable operations could encompass far more tha.n those things considered "merely mecha.nical" in common parlance, and indeed cou ld emulate human intelligence. The last of these papers, the only one to be published in his lifetime, appearing in the philosophy journa l lviind, has become famo us for the Turing Test and its 50 prophecy, and stands still as a flagship for confidence in the ultimate mechanizability of JVIind. But. Turing's constructive arguments for how Artificial Intelligence might be achieved are perhaps as significant as the long-term vision. Notably, his ideas encompassed both the "top-down" and the "bottom-up" ideas that were to become bitter rivals in later AI research. But it is also notable that. he did very li ttle to follow up these ideas with active research, even when he had the resources of the Manchester computer.

5 Introductory Biography 7 In 1951, 'TUring was elected to a Fellowship of the Royal Society, the citation referring to his 1936 work. This was a wat ~rs h ed year for Turing: although he had largely failed in the immediate post-war period to capitalize on his wartime achievement, he now sta.rted a quite fresh development, demonstrating the part he could still have in the great expansion of science and mathematics that began in the 1950s. His new ambition was that of giving a mathematical explanation for morphogenetic phenomena, thus showing an interest in biology that went back to ch ildhood, but which was now expressed in advanced methods for studying nonlinear partial difrerential equations with the computer simulations which had just become possible on the Ivlanchester computer. At the end of 1951 Turing submitted a first paper on this work, which for mathematical biology wa.s to be as important a::; hi::; 1936 work had been for logic. But at just this point, Alan Turing was arrestee!. A"" a homo""exual, he was always in danger from the law which at that time criminalizee! all homosexual activity: an injudicious liaison turned that potentia.! into fact. The trial, in March 1952, resulted in his being forced to accept injection::; of oestrogen. He fought hard to prevent thi"" from a.rre""ting his work. Unrepentant, open and unashamed, Alan Turing found himself a very isolated figure at Manchester. In 1953 there was another "crisis" with the police, which may \vell have been related to the fact. thai as a known homosexua.! he fell into the new category of "security risk," one who could no longer continue the secret work he hac! previously been doing. His holiday::; abroad to less hostile clime"" would not have calmed the nerves of security officer"". Amidst t.his Cold \Var story, however, Turing also found time not only for substantial developments in hi "" morphogenetic theory, but for a stab at a new field: the interpretation of the quantum mechanics that had first absorbed him in youth. All this was, however, cut oh by his death by cyanide poisoning at his home at \Vilmslow, Cheshire, in 1954, by means mo""t likely contrived by him to allow those who wished to do so to believe it an accid ent. An awkward figure, who delighted yet often infurintf'd hi "" friends, Alan Turing was wrapped up in world events and yet most concerned with an intense personal integrity. Writing as plainly as he spoke, he was an Orwell of science; but his large capacity for frivolity, as illustrated in hi"" discussion of the 1\lring Test setting, gave him an honorable place ill the lighter and cheekier side of English culture. His life was full of paradox, not least that he, of all people original and socially nonconforming, should be the foremost advocate of the view that the mind was purely mechanical. The most purely scientific in spirit, his application to war work was of greater effect than perhaps any other individual scientist. Committed t.o honesty and truth, he found his life enveloped by secrecy and silence. The strange drama of Alan Turing's death in 1954 has ill its way given him a lasting life in public consciousness. His state of mind at death remains an enigma, but so too does the true inner story of his life. Prickly and proud, yet

6 8 Andrew Hodges self-effacing, Thring wrote little about the development of his ideas. There is the unknown background to his fascination \. : ~ h the problem of Mind, where only juvenile fragments survive. There is t he question ra ised by Newman, of whether he might have done greater things in mathemat ics, but for the war; and the question of the rea.l motivations for Turing's aba ndonment of deep mathematical work for the sake of the war. The vexed question of the emergence of the digita.l co mputer in 1945, and of Thring's relationship with von Neuma nn, remains a ga.p at the heart of 20 th -cent.ury t echnology. The true genesis of his Artificial Intelligence program during t.he war, and the question of whether his concern for the significance of Codel's theorem was really resolved - all this remains unknown, spur to 21st-cent.ury thought and our fascination 'with the theory and practice of intelligent life. References 1. Agar, J. (2001). Turing and the Uni ve rsal fviachine (Cambridge: Icon). 2. Dav is,!'vi. (2000). Tire Universal Computer (New York: '..;orton). 3. Hodges, A. (1983). Alan Turing: the enigma (Burnett, London; Simon & Schuster, New York; new editions Vintage, London, 1992, 'Walker, New York, 2000). Further material is on http: / / org. uk. 4. Hodges, A. (1997). Turing, a natural philosopher (Phoenix, London ; ROulI('dge, New York, 1999). Included in: T he Great Philoso phers: e d~. R. Monk and F. Raphael (Weidenfeld and Nicolson, 2000). 5. Ho<igc~, A. (2002). Ala n M. Turing, in E. N. Za lta (ed.), Stallford E ncyclopedia of Philosophy, edu. 6. Newman, Iv!. H. A. (1033). Ala n JVI. Turing, Biographical memoirs of the Royal Society, Turing, A. JV!. (1992,2001). Collected Works: eds. J. L. Britton, R. O. Gandy, D. C. Ince, P. T. Saunders, C. E. 1'1. Yates (Amsterdam: North-Holland ). 8. Turing, E. S. (1959). Alan td. Turing (Cambridge IldltT!;). 9. The Turing Digital Archive a t. org off,'rs an online version of the Turing archive of papers a t. Ki ng's College, Cambrid:',c.