Open Research Onlineoro.open.ac.uk/5609/1/01299654.pdf · work. In the order of their broadcasts, they were Douglas Hartree, Max Newman, Alan Turing, ... Five 1951 BBC Broadcasts

Open Research OnlineThe Open University’s repository of research publicationsand other research outputs

Five 1951 BBC Broadcasts on Automatic CalculatingMachinesJournal ItemHow to cite:

Jones, Allan (2004). Five 1951 BBC Broadcasts on Automatic Calculating Machines. IEEE Annals of theHistory of Computing, 26(2) pp. 3–15.

For guidance on citations see FAQs.

c© [not recorded]

Version: [not recorded]

Link(s) to article on publisher’s website:http://dx.doi.org/doi:10.1109/MAHC.2004.1299654

Copyright and Moral Rights for the articles on this site are retained by the individual authors and/or other copyrightowners. For more information on Open Research Online’s data policy on reuse of materials please consult the policiespage.

oro.open.ac.uk

IEEE Annals of the History of Computing Published by the IEEE Computer Society 1058-6180/04/$20.00 © 2004 IEEE 3

Researchers of Britain’s early postwar history ofcomputing have known for some time that aseries of five British Broadcasting Corporation(BBC) radio broadcasts under the general title of“Automatic Calculating Machines” was broad-cast on the BBC’s Third Programme radio serv-ice in May–June 1951. In these broadcasts, fiveBritish pioneers of computing spoke about theirwork. In the order of their broadcasts, they wereDouglas Hartree, Max Newman, Alan Turing,Frederic (“Freddie”) Williams, and MauriceWilkes. Apart from Turing’s broadcast, whichhas been discussed by B. Jack Copeland1 andAndrew Hodges,2 these broadcasts have receivedlittle attention from historians of computing.

No sound recordings of the broadcasts sur-vive, although they all were recorded onacetate phonograph discs prior to transmission.However, texts of all five broadcasts survive asBBC transcripts, which were taken from therecordings shortly after they were made. Thesetranscripts are held at the BBC’s WrittenArchives Centre in Caversham, near Reading,and are the basis for this article.

In addition to the existence of the five BBCtranscripts, three of the speakers’ scripts areknown to have survived. These are Turing’s, heldat the Alan Turing archive at King’s College,Cambridge, and those of Wilkes and Newman,copies of which are held by Wilkes. Turing’sscript has been published,3 although curiouslynot in the Collected Works of A.M. Turing,4 and isalso available on the Word Wide Web.5 None ofthe other scripts has been published.

All five of the speakers in this series were, orhad been, involved with one or more of thethree major computing projects in the UK inthe immediate postwar period:

• ACE (Automatic Computing Engine), at theNational Physical Laboratory, designed byTuring, launched in 1946 and experimen-tally operational in a pilot version in 1950,although not completed until late 1951.6

• EDSAC (Electronic Delay Storage AutomaticComputer), at Cambridge University, designedby Wilkes, begun in 1947 and operational inMay 1949.7

• Mark 1 Prototype at Manchester University,associated with Newman, Williams, and(from 1948) Turing. Operational from April1949 to August 1950, having evolved froman earlier “Baby” test machine (operationalJune 1948) and replaced in February 1951by the Ferranti Mark 1.8

Table 1 (see p. 4) gives the titles and broad-cast dates of the talks, and the computers thatthe speakers were associated with at the time ofthe broadcasts.

As Table 1 shows, the Cambridge andManchester projects were well represented inthe five broadcasts. The National PhysicalLaboratory’s ACE computer project was repre-sented only indirectly, via Turing, who was nolonger associated with it when he made hisbroadcast. This machine was not, in any case,fully completed at the time of these broadcasts.

The Third ProgrammeBefore discussing the content of the broadcasts,

I should mention the BBC’s Third Programme, on which these five talks were transmitted. Thisidiosyncratic radio service—so unlike almostanything in modern-day broadcasting—occu-pied an important position in Britain’s intellec-

Five 1951 BBC Broadcasts onAutomatic Calculating MachinesAllan JonesOpen University

In May and June 1951, five leading figures of British computing—Douglas Hartree, Max Newman, Alan Turing, Frederic (“Freddie”)Williams, and Maurice Wilkes—spoke about their work on BBC radio.This article examines surviving texts of their broadcasts, and thespeakers’ principal points are summarized through quotations andcommentary. The broadcasts are placed in the context ofcontemporary developments in computing and the particular BBCservice on which they were broadcast.

tual life, and an appreciation of its philosophysheds useful light not only on the broadcasts buton the nature and size of the audience thatwould, or could, have heard them.

The Third Programme was a nationaldomestic radio service inaugurated by the BBCin September 1946 with an avowedly intellec-tual and cultural character. Two other nationaldomestic radio services, the Home Service andthe Light Programme, already existed—hencethe name Third Programme. Central to theactivities of the Third Programme were broad-casts of serious music, literature, and speech.Many leading thinkers of the day were invitedto speak on the service, and in general the styleof presentation was for the speaker to deliver ascripted talk typically lasting around 30 min-utes. (All the talks in this series lasted about 20minutes.) Interview-style presentations wereunusual, although widely used on the BBC’sHome Service. The fact that the five broadcastswere made by the computing pioneers them-selves, rather than by journalists or commen-tators, is thus typical of the approach used onthe Third Programme and is what makes themparticularly interesting as historical sources.

The Third Programme had no regulartimetable of program “slots”—there was no reg-ular time of the week for science broadcasts,poetry, or anything else. The only way for lis-teners to find out about forthcoming broadcastswas to consult the program listings publisheddaily in newspapers or the BBC’s own weeklyRadio Times. The five computer broadcasts dis-cussed here, therefore, did not form part of aregular science and technology stream, andwere not even broadcast at equal intervals orregular times. They nevertheless were conceivedof and presented as a series, and at the end ofeach broadcast there was an announcement ofwhen the next would take place.

The Third Programme operated only duringthe evening, and listeners were not expected tospend the whole evening listening to the serv-ice. Indeed, it was considered undesirable forthem to do so. Rather, listeners were expected to

tune in for just the broadcasts that interestedthem or which aroused their curiosity, and thenswitch off, or listen to another station. As theBBC’s historian Asa Briggs noted, “The ThirdProgramme set out not to meet the wishes of lis-teners who would be engaged in continuous lis-tening but rather to recruit ‘patrons’.”9

How many patrons the Third Programmehad at the time of these broadcasts is hard topin down. In the late 1940s, it was claimed tobe between 1.5 million and 2.5 million.9 A typ-ical audience for any single Third Programmebroadcast would naturally have been muchsmaller than this. In 1949, two years beforethese five broadcasts, the audience for a ThirdProgramme broadcast was estimated to bearound 90,000, and it appears not to havegrown during the next few years.10 Indeed, thepercentage of BBC radio listeners tuning in tothe Third Programme was generally 1 percentof the total radio audience during the early1950s.11 The more popularly oriented HomeService and the Light Programme would typi-cally have audience figures of a few million fortheir more popular broadcasts. The ThirdProgramme was subject, in any case, to techni-cal constraints that restricted its coverage to themore populous parts of the UK. Reception inmany parts of the country was poor, and inremote areas nonexistent.

Though small in absolute terms, the ThirdProgramme’s audience was nevertheless influ-ential, as Britain’s academics, artists, and intel-ligentsia were disproportionately representedamong it. However, professional intellectualswere by no means the Third Programme’s onlylisteners. A 1949 survey reported that 35 per-cent of the audience was working class,although it appears that working class was thendefined more widely, and middle class more nar-rowly, than would now be the case.12

Radio broadcasts relating to computers,cybernetics, and artificial intelligence (as wewould now call it) were by no means rare onthe BBC in this period. Between 1946 andDecember 1956, there were 24 such broadcasts,

4 IEEE Annals of the History of Computing

Five 1951 BBC Broadcasts

Table 1. “Automatic Calculating Machines” broadcasts.

Computers associated Broadcast Repeat with at time

date date Speaker Title of broadcast5 May 1951 24 June 1951 Douglas Hartree “Automatic Calculating Machines” Cambridge, EDSAC8 May 1951 26 June 1951 Max Newman “Automatic Calculating Machines” Manchester Mark 1 (Ferranti)15 May 1951 3 July 1951 Alan Turing “Can Digital Computers Think?” Manchester Mark 1 (Ferranti)2 June 1951 4 July 1951 Freddie Williams “Automatic Calculating Machines” Manchester Mark 1 (Ferranti)5 June 1951 10 July 1951 Maurice Wilkes “The Use of Automatic Calculating Machines” Cambridge, EDSAC

not counting repeats. Most of these broadcastsdate from after 1950, and most were on theThird Programme. Speakers in these otherbroadcasts included Norbert Wiener, ColinCherry, Wolfe Mays, Frank H. George, andChristopher Strachey.13 The extent of this cov-erage of computer-related matters is perhapssurprising given the widespread perception inthe UK that press and broadcasting personnelare biased against (and ignorant of) science andtechnology. However, computer-related broad-casts on the Third Programme were probably aspecial case owing to the particular interests oftheir producer, to whom I will later return.

The broadcastsAlthough much of the Third Programme’s

output was broadly educational, the ThirdProgramme was not part of the BBC’s educa-tional service. Third Programme broadcasts weretherefore not didactic in the usual sense, andspeakers were encouraged to address the listeneras an equal who just happened not to be con-versant with the speaker’s subject. Accordingly,none of the speakers in this series pitched histalk at a high technical level, and none adoptedthe style of a formal academic lecture where onemight expect a progression of ideas from funda-mentals to higher-level concepts. In this respect,the style of these broadcasts was similar to thatused for other factual talks (not just sciencetalks) on the Third Programme at that time. Ingeneral, the speakers confined themselves tofairly simple factual accounts of what comput-ers were and what they did. As so often, howev-er, Turing was something of an exception. Hispresentation, although not of a high technicallevel, certainly made greater demands on the lis-teners’ comprehension.

With the exception of Williams, the speak-ers said relatively little about the hardware,concentrating instead on software conceptssuch as programs, data, subroutines, and so on,and also touched on the recurring theme ofwhat a program in principle could and couldnot do. As regards the prehistory of computing,no speaker referred to wartime code-breakingactivities, although Williams did mention theimportance of wartime radar research for thedevelopment of computers. The names ofCharles Babbage and Lady Lovelace (that is,Ada Byron, mathematician and associate ofBabbage) are occasionally invoked as importantpioneers, but those of John von Neumann, J.Presper Eckert Jr., and John Mauchly are notmentioned at all. Their absence was probablymore out of consideration for the listener, towhom those names would have meant little,

than out of chauvinism. Hartree and Wilkes, inparticular, were happy to pay tribute to theseAmerican pioneers in their writings.

It cannot be claimed that the broadcasts sig-nificantly change our view of the history ofcomputing. The transcripts of them do never-theless offer a valuable insight into the rela-tionship of the then emerging field ofcomputer technology to the public under-standing of that technology, as revealedthrough the mouths of its leading British prac-titioners. It is against this background that thebroadcasts are most profitably viewed. Throughthe broadcasts we get a sense of what the speak-ers thought was significant in their work, whatmight be comprehensible to a nonspecialistaudience, and where developments might lead.Significantly, we also get repeated reassurancesabout where the work was not likely to lead—toward the “electronic brains” so frequentlyinvoked in popular journalism of the time.Once again, however, Turing was something ofan exception.

In the space of this article, it is impossible todiscuss each broadcast in depth. In the follow-ing five sections, therefore, I summarize eachbroadcast through quotations and commen-tary, taking the broadcasts in the order inwhich they were made. Because the transcriptswere made by nonspecialist clerical staff, thereare occasionally places where the transcriberhas clearly misinterpreted what the speaker hassaid. In my quotations I have corrected suchmisinterpretations without comment, and in afew places I have adjusted the punctuation tosomething more appropriate for a written pres-entation. Occasional interpolations of my ownare enclosed within square brackets.

Douglas HartreeThe first speaker in the series, Douglas

Hartree, had broadcast about computers on theBBC five years earlier, in December 1946, onthe Home Service. In his earlier broadcast, hehad mainly been concerned with the ENIACmachine, which he had recently used during avisit to the US.

At the time of his 1951 broadcast, Hartreewas Plummer Professor of MathematicalPhysics at Cambridge University, although inthe immediate prewar period he had been asso-ciated with developments in analog computingat Manchester University, particularly the dif-ferential analyzer.14 His inaugural lecture atCambridge had been titled “CalculatingMachines: Recent and Prospective Develop-ments,” and he had already published variouswritings relating to digital computers, notably

April–June 2004 5

his account of the ENIAC machine in Nature15

and in his book Calculating Instruments andMachines.16

In his May 1951 broadcast, Hartree was con-cerned at a basic level with the differencesbetween computers and other sorts of machine.He described the parts of a computer, the rela-tionship between data and information interms of the computer’s operation, and thetasks computers could be made to do (such ascalculating, playing games, and other appar-ently human-like activities). He began byemphasizing three salient points about themachines that were to be the subject of thisseries of talks: that they were automatic, gener-al purpose, and digital. Only the first two ofthese three points were elucidated:

By an “automatic machine” is meant one whichcan carry out numerical calculations of anylength without the attention of an operator,once the schedule of operations to be carried outhas been supplied to the machine, in a suitableform; and by a “general-purpose machine” ismeant one which can be used for a largenumber of different kinds of calculations, bysupplying it with the appropriate schedules ofoperating instructions.

The third of Hartree’s introductory points, thedigital nature of computers, was not expanded(although Newman enlarged on it in the sec-ond broadcast).

The concept of a general-purpose machinecan be traced back to Babbage’s proposed ana-lytical engine. Hartree was aware of Babbage’swork and mentioned it in passing as represent-ing the first conception of a general-purposedigital machine. He then launched into theanatomy of the modern (that is, vonNeumann) machine:

An automatic digital calculating machine con-sists of five main parts, an arithmetical unit, astore, a control unit, an input unit, and an out-put unit. The purpose of the store is to holdinformation, either numbers or operatinginstructions, for as long as they may be required,in the course of the calculation. In some of theolder machines, the store consisted of two dis-tinct parts, one for numbers and one for instruc-tions. But in most of the more recent machines,the same store is used both for numbers and forinstructions.

By “the older machines” Hartree was refer-ring to machines such as the AutomaticSequence Controller at Harvard University, in

which data was held on counters and instruc-tions on punched paper tape that was read asthe calculation proceeded, or the ENIAC, inwhich the “program” was assembled physicallyby setting switches and by patching togetherprocessing units via plugboards and cables. Themore modern machines not only held instruc-tions and data in the same memory, but madeno distinction in the way they were held:

But in most of the recent machines there is nodistinction between the form used for numbersand for instructions. The distinction betweenwords representing numbers and words repre-senting instructions lies in the way in which theyare used.

A consequence of this lack of distinctionbetween data and instructions is the possibili-ty of self-modifying programs (something onwhich more than one speaker was to com-ment):

This possibility of modifying instructions as thecalculation proceeds provides the means ofinstructing the machine to carry out much of thediscrimination and selection between alternativeprocedures which a human computer wouldexercise in doing the same calculation by penciland paper methods.

Hartree raised here the contentious issue of theanalogy between humans and computers.From the announcement of the ACE project inautumn 1946 (the first of the British comput-ing projects to be announced publicly), thepress had had a tendency to refer to the newcomputers as “brains,” or “electronic brains.”17

Hartree was anxious to correct what he viewedas a misapprehension:

But do not jump to the conclusion that [in mod-ifying its own program] the machine is thinkingfor itself. All these instructions for modifyingother instructions, and for evaluating and usingthe criteria of any discrimination, have to bethought out and programmed in detail. Themachine only carries out literally and blindlyand without thinking, the instructions whichthe programmer has thought out for it.

Turing, later in the series, took a differentview, as we shall see.

Martin Campbell-Kelly18 has written thatone of the distinctive features of computing asdone at Cambridge at this time was the empha-sis on building up a library of commonly usedsubroutines. Hartree alluded to this policy:

6 IEEE Annals of the History of Computing

Five 1951 BBC Broadcasts

… the machine and the process of providing itwith instructions may be such that groups ofoperations for standard processes, such as theevaluation of square roots and cube roots, can beprogrammed once and for all. […] The mainwork in the preparation of the calculations forthe machine may then be the programming of amaster routine consisting mainly of instructionsfor calling in the subroutines in the propersequence.

Hartree concluded by mentioning the pos-sibility of computers playing games such aschess, but he would not regard this as evidenceof thinking:

… [playing games] would come very near what,in ordinary speech, we would call thinking—anaspect of those machines on which I understandDr. Turing will be speaking. But remember, thatthe sequence of operations for such a process stillhas to be programmed, and Lady Lovelace’swords still apply—“the machine can only dowhat we know how to order it to perform.”

This remark of Lady Lovelace’s is a recurringtheme in the first three broadcasts, with eachspeaker giving a different verdict on its veracity.

Max NewmanThe mathematician Max Newman had

worked on breaking German Enigma–codedmessages at Bletchley Park during World War IIalong with Turing, who had been one of hisstudents at Cambridge University before thewar. At the time of his 1951 broadcast,Newman was professor of mathematics atManchester University, and his interest in com-puting was mainly with a view to their use as amathematics research tool. Although Newmanhad largely initiated the project to build a com-puter at Manchester, he had little involvementwith the design of its hardware.19

In his talk, Newman was concerned withwhat made computers so powerful. He locatedtheir power in the fact that they used a limitedset of elementary operations, and any sequenceof operations could be repeated until a stipu-lated condition was satisfied.

At the start of his talk, he picked upHartree’s idea of the general-purpose machine,which could perform a wide range of tasksdespite its relatively small repertoire of ele-mentary operations:

It is the arrangement of these elementary opera-tions, and the way they are interrelated, thatcauses us to call one series [that is, program] a

way of solving equations, and another a routinefor playing bridge. [….] Problems that appear notto be arithmetical at all may often be made so, byquite trivial changes in the way they are stated.

However, the existence of a set of elementaryoperations is not by itself what gives a com-puter its power:

If an automatic computing machine really need-ed a tape containing 100,000 instructions inorder to do 100,000 elementary operations,somebody would have to punch the tape; andthat “somebody” might be just as usefullyemployed in doing the 100,000 elementary sumshimself, with a pencil and a piece of paper.

This is perhaps debatable. Even if a task with100,000 operations required a program tapecontaining 100,000 instructions, there mightstill be a benefit in creating the tape becausethe program could be used many times toprocess different sets of data. As far as Newmanwas concerned, however, the utility of a com-puter lay in the fact that a multistep operationcan be specified in fewer steps than the opera-tion itself would take:

The machines that are the subject of this talk […]all have the essential property of being able to doa big job from a few instructions. … [The]arrangements by which this is achieved are themost characteristic feature of these machines,and are the source of those complexities ofbehaviour that give some colour to comparisonswith certain mental processes, […].

Newman went on to mention the “jump”instruction as one technique for doing “a bigjob from few instructions,” by enabling asequence of operations to be repeated:

The normal procedure, when the machine is start-ed off, is for the instruction in line 1 to be carriedout first; then control passes to line 2, the instruc-tion in it is carried out, control passes on to line3, and so on. … [There] is a special type of instruc-tion whose function is precisely to interrupt thenormal succession. For example, Instruction 100might be “Jump back to instruction 25.”

Of course, one needs to be able to exit fromthe loop created by jumping back to an earlierinstruction:

There must be some way of bringing the repeti-tions of a cycle to an end when they have goneon long enough. […] This is accomplished by

April–June 2004 7

introducing a conditional or branched type ofinstruction, for example: “If line 27 is empty (i.e.,contains 0) step on to the next instruction in thenormal way; but if it is not empty jump toinstruction 12.”

Such accounts of conditional jumps have away of being couched in anthropomorphicterms, which Newman wished to counter:

Now there is some danger here that the jargon of“obeying instructions” and “choosing alterna-tives” which has become the customary way ofdescribing the behaviour of these machines, mayevoke a picture of the machine “conning” [thatis, reading and memorizing] the branchedinstruction, looking to see if line 27 is empty, andthen faithfully choosing the appointed alterna-tive. In fact the machine “obeys” its instructionsin exactly the same sense that a railway train“obeys” the points [that is, switches], going toCrewe if they are set one way and to Macclesfieldif the other.

Presumably, Newman intended his railwayanalogy to suggest that a program can no morevary its route during a calculation than can atrain driver during a journey: Once a programand its data are read into a computer, the futurecourse of the data-processing operation is ascompletely determined as is the route of atrain. However, the analogy is potentially mis-leading. The course of a train is knowable inadvance, but this is not necessarily true of acomputing program’s calculations, as Newmanhimself said later (quoted below).

Like Hartree, Newman saw self-modifyingprograms as holding an intriguing possibilityfor something close to what we would now callartificial intelligence:

The machine will add lines 2 and 3, if instructedto do so, without the least regard to whether oneor the other of these lines is to be used later on asan instruction. This means that we can modifynot only the true numerical material, but also theinstructions themselves, in the course of the com-putation. […] This has, with some justification,been described as the ability to learn from results.

However, whereas Hartree was clear thatsuch self-modifying programs could only “lit-erally and blindly” carry out the programmer’sinstructions, Newman was less certain:

It is not difficult to make up programmes ofmoderate length leading to networks of opera-tions so complex that even the composer [that

is, programmer] cannot predict what course thecalculations will take, and it is not obvious thatanyone could discover a routine to obtain theresults of such a programme […]. In view of thesefacts it seems that the dictum of Lady Lovelace,as quoted by Professor Hartree, that “themachine can only do what we know how toorder it to perform,” needs to be received withsome reserve. However the end of my talk is notthe place to enter on these fascinating but con-troversial topics.

Alan TuringBy the time of Turing’s broadcast, roughly a

year had passed since the publication of hisnow famous Mind article in which he discussedthe issue of whether computers could be said tothink.20 At the outset of his broadcast, Turingmade it clear where he stood:

Digital computers have often been described asmechanical brains. Most scientists probablyregard this description as a mere newspaperstunt, but some do not. One mathematician hasexpressed the opposite point of view to me ratherforcefully in the words “It is commonly said thatthese machines are not brains, but you and Iknow that they are.” […] I shall give most atten-tion to the view which I hold myself, that it isnot altogether unreasonable to describe digitalcomputers as brains.

Much of the rest of the talk is a summary ofTuring’s justification for regarding computerspotentially as brains, and the kinds of reasonthat people put forward to oppose the sugges-tion that computers might one day be able tothink. Of these objections, the principal one isLady Lovelace’s argument that computers onlydo what they have been programmed to do.

Turing was careful to make clear that thecomputers of his day could not plausibly becalled brains; his point is that digital comput-ers had the potential for being plausibly regard-ed as brains. His argument, familiar from theMind article, depends on the concept of theuniversal machine, which he had conceived inconnection with his celebrated 1936 paper,21

although he did not mention that paper here:

A digital computer is a universal machine in thesense that it can be made to replace anymachine of a certain very wide class. It will notreplace a bulldozer or a steam-engine or a tele-scope, but it will replace any rival design of cal-culating machine, that is to say any machineinto which one can feed data and which willlater print out results.

8 IEEE Annals of the History of Computing

Five 1951 BBC Broadcasts

The next step of Turing’s argument dependson a view of the brain that remains controver-sial, although it has a long ancestry in materi-alist philosophy:

If it is accepted that real brains, as found in ani-mals, and in particular in men, are a sort ofmachine, it will follow that our digital comput-er, suitably programmed, will behave like a brain.

The success or otherwise of this emulationis to be assessed in the test that now bearsTuring’s name, which he explained in his Mindpaper and which he summarized here briefly:

I think it is probable for instance that at the end ofthe century it will be possible to programme amachine to answer questions in such a way that itwill be extremely difficult to guess whether theanswers are being given by a man or by themachine. I am imagining something like a viva-voce examination, but with the questions andanswers all typewritten in order that we need notconsider such irrelevant matters as the faithfulnesswith which the human voice can be imitated.

Allowing the computer to respond non-oral-ly, via a typewriting machine, suggests thatTuring thought the problem of programming acomputer to speak convincingly was even morechallenging than that of programming it torespond plausibly to questions. Turing’s ownwork on speech sampling and encipherment,in the latter part of World War II, may have ledhim to view speech synthesis as a particularlyintractable problem.

Given Turing’s view of the potentially brain-like behavior of computers, it is perhaps no sur-prise that he considered they might one day becapable of originality:

If we give the machine a programme whichresults in its doing something interesting whichwe had not anticipated, I should be inclined tosay that the machine had originated something,rather than to claim that its behaviour wasimplicit in the programme, and therefore thatthe originality lies entirely with us.

Turing acknowledged that there areimmense difficulties to be overcome before acomputer could behave in a convincinglyhuman way, nor did he know how one wouldgo about programming a machine to behave insuch a way:

I will only say this, that I believe the processshould bear a close relation to that of teaching.

The essential point of Turing’s observationson the difficulty of programming brainlikebehavior is that the programmer may notalways know what the consequences of a pro-gram may be:

Let us now reconsider Lady Lovelace’s dictum.“The machine can do whatever we know how toorder it to perform.” The sense of the rest of thepassage is such that one is tempted to say thatthe machine can only do what we know how toorder it to perform. But I think this would not betrue. Certainly the machine can only do what wedo order it to perform, anything else would be amechanical fault. But there is no need to supposethat, when we give it its orders, we know whatwe are doing, what the consequences of theseorders are going to be.

Thus whereas Hartree accepted the Lovelacedictum and Newman felt reservations about it,Turing rejected it—or at any rate rejected whatit is usually taken to imply. For Turing, the factthat a program together with its data is a deter-ministic system (that is, its initial state fullydetermines its route to its final state) does notpreclude brainlike or original behavior becausewe cannot necessarily predict what that finalstate will be and how it is reached.

Turing’s talk was the only one of the seriesfor which I have been able to find a review.Writing in The Listener, a weekly BBC publica-tion, Martin Armstrong wrote:22

… I was moved every few minutes to hold up myschoolboy hand with a “Please, Sir … Onemoment, Sir …. Will you explain what you mean,Sir, by …” this that and the other. Mr Turingremarked that many people dislike the idea thata machine could be made to think. “If machinescould think,” they say, “where would we be?

Now I, as it happens, am one of those who dis-like the idea, not, however, because it frightensme, but because it seems to me to be based on amisuse of words. To say that a machine thinks issurely, by implication, to define thought as amechanical process, …

One can sympathize with the reviewer’s dif-ficulties in following Turing’s talk, which wascertainly the densest of the series. To havegrasped all Turing’s points on a single hearingwould not have been easy.

Regarding the definition of “thinking” or“thought,” which the reviewer had troublewith, Turing did actually give a sort of defini-tion in his talk, although it would have beeneasy to miss:

April–June 2004 9

… to programme a machine to imitate a brain,or as we might say more briefly, if less accurate-ly, to think.

Turing’s comment here explicitly, if approx-imately, equates imitating a human brain withthinking, which is rather different from whatwe find in his Mind article. In that article,Turing stated that he did not want to be drawninto defining “thinking,” and proposed his testas a way of avoiding the ambiguities associatedwith the word. What the test actually tests isnot stated explicitly in the Mind article,although commentators have usually inter-preted it as an operational test of eithermachine-based intelligence or machine-basedthought. The informality of this radio presen-tation, however, appears to have encouragedTuring to use the terms “thought” and “think-ing” in connection with machines rather moreopenly than he did in more formal contexts.Here are two further examples:

I will not attempt to say much about how thisprocess of “programming a machine to think” isto be done.

I have tried to explain what are the mainrational arguments for and against the theorythat machines could be made to think …

Earlier, I mentioned that Turing’s broadcasthas been discussed by Hodges and Copeland.Hodges’s discussion is confined to a few sen-tences pointing out that Turing reiterated in hisbroadcast ideas he had already expounded else-where, principally in the Mind article. This istrue, but the broadcast at least presented a con-cise summary of those ideas, aimed at a nonspe-cialist audience. Copeland’s discussion is muchlonger, and for the most part is less concernedwith the broadcast than with the interpretationof the phrase “any machine” (taken from thebroadcast). Copeland devotes a few paragraphsto Turing’s suggestion that the appearance of freewill in a computer may be created by the inclu-sion of a random process in the program, butthis portion of Turing’s talk was relatively short.

Freddie WilliamsFreddie Williams’s down-to-earth talk could

hardly have been in greater contrast to Turing’s.Williams was an electrical engineer who, alongwith Tom Kilburn, had devised the highlyinnovative cathode-ray tube memory used inthe Manchester Baby machine, the Mark I pro-totype, and the Ferranti Mark I (installedFebruary 1951).23 His talk was almost entirelyconcerned with the principles of computer

memory, of which—leaving aside electronicmemory—there were really only two practicaltypes at the time of the broadcasts: mercurydelay lines and cathode-ray tube devices.

After briefly describing the power of com-puters to do large-scale calculations at highspeed, Williams outlined the engineering prob-lem faced by the designer of computer memory:

Thus the problem reduces to finding somewhereto put strings of 0’s and 1’s, about half a million ofthem altogether, and it must be somewhere wherethey can be got at in sets of 20 or 40 within say athousandth of a second when they are wanted.

The Cambridge EDSAC machine and theACE machine used batteries of mercury delaylines for their memory. These were tubes ofmercury down which pressure waves weretransmitted. The operation of these devices waslikened by Williams to a man shouting to a dis-tant cliff face:

If he shouted a number and then listened, after acertain time he would hear an echo. He could thenshout the number again, the only tax on his mem-ory would be between hearing an echo and shout-ing again; after a further delay he would again hearan echo and shout, and so keep the number circu-lating between himself and the cliff. If the echowere delayed a longish time he could shout sever-al numbers before the first one came back and sokeep several numbers in circulation. Thus oneman with a poor memory could store a lot of num-bers in the air, just by repeating what he heard.

Williams acknowledged that this analogy isonly appropriate up to a point. In a practicalmercury tube, the data (in the form of ultra-sonic pressure waves) are not echoed back tothe transmitter along the tube. Instead, theyare received by a transducer at the far end ofthe tube which

… returns the signals to the near end electrical-ly, where they are regenerated and retimed rela-tive to the clock by an electrical circuit.

The cathode-ray tube memory, in contrast,arranges the data two-dimensionally in space:

Imagine a man supplied with a square of dustsubdivided by low walls into a lot of little squares,just like an egg box, and supplied also with a stickto scratch the dust. Now let him be read a row of0’s and 1’s and let him start at the top left-handbox and progress from box to box from left toright and top to bottom, as in reading, writing

10 IEEE Annals of the History of Computing

Five 1951 BBC Broadcasts

“1” when we say “1” and “0” when we say “0”.In the electrical version the man, the stick,

and the square of dust are replaced by a cathode-ray tube […] The square of dust is the face of thecathode-ray tube, the dust on it is the inex-haustible supply of secondary electrons that canbe knocked out of it by a high velocity primaryelectron beam. This beam is itself used as thestick, and it is moved about in much the sameway as in television …

The electron beam not only wrote data intothe memory but also read it. By scanning thescreen, the beam could detect excited andunexcited regions of the screen, as in a televi-sion camera of the sort used in those days.

A snag with this sort of memory was thatdata in the memory tended to be corrupted bywhat Williams referred to as the “splashing ofsecondary electrons from one box to its neigh-bours,” and therefore the image on the screenneeded regular regeneration:

We have found that over a thousand separate“boxes” can be set up on one cathode-ray tubebefore the “splashing” from box to box makesthe fading between regenerative visits of beamtoo great to be tolerated.

William’s cathode-ray tube memory was aformidable engineering achievement and sim-ilar types of memory were used not only in theUK (in early Ferranti machines) but also in theUS, where it was used in, for instance, theWhirlwind machine at the MassachusettsInstitute of Technology during its early years.Nevertheless, computer memory remained anawkward and unreliable technology:

But neither [mercury delay-line nor cathode-ray-tube memory] really solves the whole problem,since to store 500,000 digits would require noless than 500 [cathode-ray] tubes or [mercury]delay lines. We have progressed beyond thispoint by using in conjunction with cathode-raytubes secondary methods of storage which lackthe property of extremely rapid access to indi-vidual numbers.

Possibly Williams was here alluding to theinnovative magnetic-drum storage used atManchester. Even this combination of cathode-ray tube memory with other forms of storageleft much to be desired:

But the ideal has not been reached—in fact onemay well conclude by saying, “the researchcontinues.”

Within a year or two, research into comput-er memory (mainly in the US) yielded a newtype that, as Wilkes put it in a September 2001interview with me, rapidly transformed thememory from being the least reliable part of acomputer into being the most reliable. This wasmagnetic-core memory, which used arrays ofthousands of small magnetic rings threaded oncurrent-carrying wires. However, practicalapplications of such memory were still in thefuture at the time of these broadcasts.

Maurice WilkesThe final speaker in the series, Maurice

Wilkes, was at the time of the broadcasts run-ning what was probably the most successful ofthe three British computer ventures, not onlyfrom the technical point of view but also fromthe organizational point of view. By mid-1951the Cambridge EDSAC, designed by Wilkes,was not just a functioning laboratory machinebut a facility used by several departments ofthe university.24

Wilkes’s talk concentrated on the scientificuse of the EDSAC machine:

Already the EDSAC has contributed to a numberof [research projects]. Astronomy and astro-physics are represented by problems connectedwith the orbits of minor planets and the equilib-rium of gaseous stars, geophysics by calculationsconcerning the propagation of wireless waves inthe ionosphere, and the effect of the motion ofa ship on a pendulum used for gravity survey atsea. The machine has also been used for statisti-cal calculations arising in applied economics andfor problems in X-ray crystallography. We areabout to start on a problem connected with thetransmission of impulses along nerve fibres.

As mentioned, a distinctive feature ofWilkes’s policy at Cambridge was the early cre-ation of a large library of subroutines:

You may be interested to know that when a high-speed electronic calculating machine is beingused, it is generally better to calculate sines andcosines afresh from a series whenever they arerequired rather than to put a table into the storeof the machine. It has also proved possible toconstruct sub-routines for carrying out some ofthe standard processes of numerical mathemat-ics, such as numerical integration, or the numer-ical solutions of differential equations. […] Thelibrary associated with the EDSAC now containsabout 150 sub-routines and it is still growing.

For comparison, the subroutine library at

April–June 2004 11

Manchester at the time of this broadcast wasabout one-third the size given here by Wilkes.25

The EDSAC library was, in fact, highly influen-tial and adopted directly by some latermachines such as the LEO machine, used bythe J. Lyons Company, and the TAC machinein Tokyo.26

Another of Wilkes’s concerns was to stream-line, and even mechanize, the business of pro-gramming:

There are a number of other tasks connected withprogramming which at first sight appear to requirethe application of human intelligence but whichcan really be done according to a set of rules.

Not surprisingly, given the large amount ofcomputing activity that had been going on atCambridge, issues of programming errors andreliability had started to loom large:

When drawing up a programme it is very easy tomake slips of a trivial kind; for example, one mayforget to make sure that the accumulator registerof the machine is cleared before beginning toadd up a series of numbers. You might think thatthese slips could all be detected by going throughthe programme carefully before putting it on themachine, but experience has shown that it is notquite as simple as that. Some of the slips are sureto get through and a good deal of the time takenin putting a new problem on the machine isspent in finding them.

Some commentators, or perhaps users, wereworried that the computer itself might intro-duce errors in calculations. Wilkes said:

… it has even been suggested that the ideal com-puting system would consist of two identicalmachines connected together in such a way thatunless they produced identical numbers at eachstage the calculation would stop. I am ratheragainst this for electronic machines at their pres-ent stage of development, mainly because themachines are quite complicated enough as it iswithout making them any more so. The EDSAC,for example, contains 3000 valves [tubes]. Themore equipment there is in a machine, the morelikely it is to go wrong.

The likelihood of a machine breaking downwas indeed high. In the EDSAC’s early days, theinterval between failures was typically a matterof minutes, although reliability steadilyimproved.27

In rounding off his talk, Wilkes was alsorounding off this series of five broadcasts:

That the future will bring important and excit-ing developments, I do not doubt, but it must beremembered that from the point of view of prac-tical achievement the subject is still in its earlystages; the number of electronic digital calculat-ing machines in operation at the present timecan be counted on the fingers of one hand, andno machine has yet been programmed to play agame of bridge.

Context of the broadcastsTo help take stock of this series of broad-

casts, it is useful to look at the wider context inwhich they took place.

The broadcasts were examples of the topicalinterest in computers coinciding with theemergence of working computers in Britainaround 1949–1950. Other manifestations ofthe same interest can be seen in the article“Can Machines Think?” that Wilkes wrote forthe Spectator around this time,28 and the articlespublished in Penguin’s Science News.29 Both ofthese publications would have been aimed atthe same audience as the Third Programmereached—that is, well-educated nonspecialists.At a more populist level, there was the Ferranticomputer displayed at the Festival of Britain(May–September 1951), which aroused a greatdeal of interest. Members of the public couldchallenge it to a game of Nim—a simple gamein which opposing players take turns to removeone, two, or three counters from an initialarrangement of 13 pieces. The winner takes thelast counter. In all these cases, the issue ofmechanical intelligence was never far away,and frequently alluded to—as is the case withthe five broadcasts discussed here. Clearly inthe lay mind, or in the minds of the peoplewho were addressing the lay mind, this issuecould not be ignored.

Aside from this rather sensationalist interest,however, the broadcasts were timely in anoth-er respect. The Cambridge and Manchestermachines were beyond being laboratory novel-ties when these broadcasts were made, and forsome time had been earning their keep as sci-entific and mathematical tools. The Manchestermachine, for instance, besides being used byTuring for his research, was available for outsideresearch projects such as the UK’s first atomicbomb project.30 As for the Cambridge machine,Wilkes’s broadcast summarized some of itsresearch uses. For some months, indeed, Wilkeshad been thinking about a replacementmachine, funds for which were being canvassedat the time of his broadcast.31

Another context in which to consider these

12 IEEE Annals of the History of Computing

Five 1951 BBC Broadcasts

broadcasts was the Third Programme itself andits relationship to other BBC radio services.(Television was still very much a minoritymedium throughout most of the 1950s in theUK.) It is at the very least surprising that theThird Programme, which carried no regular sci-ence output, carried these broadcasts and sev-eral others relating to computers, whereas theHome Service, which broadcast a weekly sci-ence program, should have had relatively fewbroadcasts on the subject (only four significantbroadcasts in the period 1950–1955). In myview, several factors account for this.

To a degree the Home Service’s science-mag-azine approach would have been more strong-ly driven by news values than was coverage onthe Third Programme, and although comput-ers were “new” during the early 1950s, theyceased to be news as they evolved from beinglaboratory projects into scientific tools. Also,the Home Service’s own news bulletins (asopposed to science broadcasts) would probablyhave covered computers at their most topicaland newsworthy moments. There was, forinstance, an item on a Home Service newsbroadcast on 9 December 1946 covering theannouncement of the ACE project. The ThirdProgramme, by contrast, was not so concernedwith the news agenda and did not even carrynews bulletins during that period. Thus, bybeing less driven by a news agenda, the ThirdProgramme’s coverage could afford to be morereflective and long term.

The most significant factor in the ThirdProgramme’s coverage, however, appears tohave been a remarkable producer of talks on theThird Programme: Theophilus Stephen Gregory.He was responsible not only for the five BBCbroadcasts but also for most of the other com-puter-related broadcasts on the ThirdProgramme during this period. Gregory was asingular character, having been a Methodistminister during the 1920s, later converting toRoman Catholicism.32 His particular areas ofinterest were philosophy and theology, and itseems probable that the contemporary debatesabout “electronic brains” caught his attentionin a way that they might not had he simplybeen a science-trained producer of sciencebroadcasts. For instance, a year before the fivebroadcasts, Gregory had produced two talksentitled “Mind-like Behaviour in Machines,”both given by Donald M. MacKay, a physicist atKing’s College, London, with a particular inter-est in the compatibility of science and religiousfaith. Later broadcasts produced by Gregory hadsuch titles as “On Comparing the Brain withMachines” (two broadcasts, again with MacKay

as the speaker), “Machines and HumanBehaviour” (with Frank H. George), and so on.Further evidence of the philosophical nature ofGregory’s interest in computers can be found ina continuity announcement that survives withthe BBC transcripts and which would have beenwritten by Gregory himself (the emphasis in thefollowing quote is mine):

This evening we are repeating the first of fivetalks on the history and theory of thinkingmechanisms ….

Whatever the particular bias of Gregory’sinterest in the subject, however, he appears tohave made no attempt to influence the broad-casters in the content of their talks, if the expe-rience of one speaker can be taken as typical.Wilkes reported in the September 2001 inter-view with me that Gregory gave him no brief-ing about his talk, and made no interventionapart from requesting him to alter the pronun-ciation of certain words (to no benefit, inWilkes’s view).

Approximately seven months after the seriesof talks discussed here was broadcast, Turingand Newman took part in another computer-related broadcast produced by Gregory on theThird Programme. This was a panel discussionentitled “Can Automatic Calculating MachinesBe Said To Think?,” broadcast on 14 January1952 and repeated on 23 January 1952.33 Otherparticipants included Geoffrey Jefferson, a pro-fessor of neurosurgery at the University ofManchester, and Richard Braithwaite, aphilosopher and Fellow of King’s College,Cambridge. However, apart from this panel dis-cussion, none of the broadcasters mentionedhere took part in any further computer-relatedbroadcasts on the BBC. Other broadcasts relat-ing to computers continued to be made, how-ever, and often by distinguished speakers.

As for the significance of these broadcasts toeither the development of computing inBritain, or to the public understanding of thesubject, these are imponderable matters pend-ing further research. However, one concreteoutcome deserves mention, even if it could notbe said to be typical. After hearing Turing’sbroadcast, enterprising amateur computerenthusiast Christopher Strachey—at the time ateacher at Harrow School—dashed off a letterto the speaker:34

Dear Turing,

I have just been listening to your talk on theThird Programme. Most stimulating and, I sus-

April–June 2004 13

pect to many people, provocative, but it fitsextraordinarily well with what I have been think-ing on the subject.

The remainder of Strachey’s four-page letterconsists of his observations on the idea thatmaking a computer think would be similar tothe process of teaching, a matter touched on inpassing by Turing and clearly related toStrachey’s professional interests as a teacher. Inclosing, Strachey wrote:

Please excuse such a long letter—I am quite sureyou are far too busy to answer it—you mustblame your talk for being far too stimulating.

Strachey’s letter thus stands as an exampleof the power of this type of broadcast to stim-ulate and illuminate. Nor was this letter theend of the matter for Strachey. He went on tobecome a remarkable theorist of computer pro-gramming and the founder, in the 1960s, of theProgramming Research Group at the OxfordUniversity Computing Laboratory.35 AlthoughI would not wish to imply that Strachey’s sub-sequent career was entirely attributable to hishearing Turing’s talk, the talk was neverthelesspart of his intellectual background and aninspiration to him. Fittingly, Strachey was him-self later to broadcast at least three times oncomputer-related matters on the BBC.

AcknowledgmentsI am grateful to Sir Maurice Wilkes for his manyhelpful comments on a draft of this article, andto Jeff Walden and the staff of the BBC WrittenArchives Centre at Caversham, Reading, UK.

References and notes1. B.J. Copeland, “A Lecture and Two Radio Broad-

casts on Machine Intelligence by Alan Turing,”Machine Intelligence, vol. 15, K. Furukawa, S.Michie, and S. Muggleton, eds., Oxford Univ.Press, 1999, pp. 445-446 and 448-453.

2. A. Hodges, Alan Turing: The Enigma, Vintage1992, pp. 441-442.

3. K. Furukawa, S. Michie, and S. Muggleton, eds.,Machine Intelligence, vol. 15, Oxford Univ. Press,1999, pp. 462-465.

4. Neither of Turing’s radio broadcasts is reprintedin the most relevant volume of his collectedworks, namely Collected Works of A.M. Turing,Mechanical Intelligence, North Holland, 1992,D.C. Ince, ed., (part of the Collected Works of A.M.Turing series).

5. The script is item AMT/B/5 in the Turing Archive,King’s College, Cambridge;http://www.turingarchive.org/browse.php/B/5.

6. M. Campbell-Kelly, “Programming the Pilot ACE:Early Programming Activity at the National Physi-cal Laboratory,” Annals of the History of Comput-ing, vol. 3, no. 2, Apr. 1981, p. 133.

7. M. Campbell-Kelly, “Programming the EDSAC:Early Programming Activity at the University ofCambridge,” Annals of the History of Computing,vol. 2, no. 1, Jan. 1980, p. 7.

8. M. Campbell-Kelly, “Programming the Mark I;Early Programming Activity at the University ofManchester,” Annals of the History of Computing,vol. 2, no. 2, Apr. 1980, pp. 130-131.

9. A. Briggs, Sound and Vision, Oxford Univ. Press,1979, p. 66.

10. H. Carpenter, The Envy of the World: Fifty Years ofthe BBC Third Programme and Radio 3,1946–1996, Weidenfeld and Nicholson, 1996,pp. 96, 109.

11. B. Paulu, British Broadcasting: Radio and Televisionin the United Kingdom, Univ. of Minnesota Press,1956, p. 368.

12. A. Briggs, Sound and Vision, p. 83.13. A. Jones, “Pioneers on the Air,” Actes du sixième

Colloque sur l’Histoire de l’Informatique et desRéseaux, Éditions ACONIT, 2002, pp. 14-28.

14. M. Croarken, Early Scientific Computing in Britain,Oxford Univ. Press, 1990, pp. 50-53.

15. D. Hartree, “The ENIAC, an Electronic Comput-ing Machine,” Nature, vol. 158, no. 4015, 12Oct. 1946, pp. 500-506.

16. D. Hartree, Calculating Instruments and Machines,Univ. of Illinois Press, 1949.

17. “An Electronic Brain,” The Times, 1 Nov. 1946, p.2; “The Mechanical Brain,” The Times, 11 June1949, p. 4; and “The Mechanical Brain,” TheTimes, 16 June 1949, p. 2.

18. M. Campbell-Kelly, “Introduction,” to The Prepa-ration of Programs for an Electronic Digital Comput-er by M. Wilkes, D. Wheeler, and S. Gill; CharlesBabbage Inst. Reprint Series, Tomash, 1982 (firstpublished by Addison-Wesley, 1951), pp. xix-xx.

19. M. Croarken, Early Scientific Computing in Britain,pp. 119-122.

20. A.M. Turing, “Computing Machinery andIntelligence,” Mind, vol. LIX, no. 236, 1950, pp.443-460.

21. A.M. Turing, “On Computable Numbers, with anApplication to the Entscheidungsproblem,” Proc.London Math. Soc., vol. 2, no. 42, 1936, pp. 230-267.

22. M. Armstrong, The Listener, 24 May 1951, p. 851.23. M. Campbell-Kelly, “Programming the Mark I;

Early Programming Activity at the University ofManchester,” Annals of the History of Computing,vol. 2, no. 2, Apr. 1980, pp. 130 and 165.

24. M.V. Wilkes, Memoirs of a Computer Pioneer, MITPress Series in the History of Computing, B.Cohen, ed., 1985, MIT Press, p. 184.

14 IEEE Annals of the History of Computing

Five 1951 BBC Broadcasts

25. M. Campbell-Kelly, “Programming the Mark I;Early Programming Activity at the University ofManchester,” Annals of the History of Computing,vol. 2, no. 2, Apr. 1980, p. 145.

26. M. Campbell-Kelly, “Introduction,” to The Prepa-ration of Programs for an Electronic Digital Comput-er by M. Wilkes, D. Wheeler, and S. Gill, p. xiv.

27. M. Campbell-Kelly, “Programming the EDSAC:Early Programming Activity at the University ofCambridge,” Annals of the History of Computing,vol. 2, no. 1, Jan. 1980, p. 12.

28. M.V. Wilkes, “Can Machines Think?,” The Specta-tor, 10 Aug. 1951, pp. 177-178.

29. “‘Memory’ in a Mercury Tube,” Science News,vol. 5, Penguin, Nov. 1947, pp. 64-66; G. Rabel,“Mathematical Instruments and CalculatingMachines,” Science News, vol. 7, Penguin, June1948, pp. 112-124; and S. Byard, “Robots WhichPlay Games,” Science News, vol. 16, Penguin,June 1950, pp. 65-77.

30. A. Hodges, Alan Turing: The Enigma, p. 440.31. M.V. Wilkes, Memoirs of a Computer Pioneer, pp.

184-188.32. H. Carpenter, The Envy of the World: Fifty Years, p.

193.33. A copy of the BBC transcript of this broadcast is

held in the Turing Archive at King’s College,Cambridge, as item AMT/B/6. The text isavailable online at http://www.turingarchive.org/browse.php/B/6 and has been published inB.J. Copeland’s “A Lecture and Two Radio Broad-casts on Machine Intelligence by Alan Turing,”Machine Intelligence, vol. 15, K. Furukawa, S.Michie, and S. Muggleton, eds., pp. 465-476.The broadcast is also discussed by Copeland hereon pp. 453-458 and by A. Hodges, Alan Turing:The Enigma, pp. 450-452.

34. Turing archive, item AMT/D/5, http://www.turingarchive.org/browse.php/D/5.

35. For a biographical note about Strachey, see M.Campbell-Kelly, “Christopher Strachey,1916–1975: A Biographical Note,” Annals of theHistory of Computing, vol. 7, no. 1, Jan. 1985, pp.19-42.

Allan Jones is a lecturer in theDepartment of Information andCommunication Technologiesat the Open University, the UK’slargest distance-teaching institu-tion. His interests include com-munication technology, musicand the history of science broad-

casting. Jones holds engineering and science degreesfrom Liverpool University and the Open University.

Readers may contact Jones at [email protected].

Models of a ManEssays in Memoryof Herbert A. Simonedited by Mie Augierand James G. March

Essays that pay tribute to the wide-ranging influence of the late HerbertSimon, by friends and colleagues.584 pp., 20 illus. $45.95

The Turing TestVerbal Behavioras the Hallmark of Intelligenceedited by Stuart Shieber

Historical and contemporary paperson the philosophical issues raised bythe Turing Test as a criterion forintelligence.A Bradford Book336 pp., 6 illus. $30 paper

now in paperback

Mechanizing ProofComputing, Risk, and TrustDonald MacKenzie“A most readable account of howprogram verification came to promiseso much and deliver so little.”— Richard Clayton, The Times HigherEducation SupplementInside Technology series440 pp., 48 illus. $28 paper

now in paperback

From AirlineReservations toSonic the HedgehogA History of the Software IndustryMartin Campbell-Kelly

“A well-rounded look at the softwareindustry from a business perspec-tive.” — Library Journal392 pp., 23 illus. $16.95 paper

To order call800-405-1619.Prices subject tochange without notice.

newfromTheMITPress

http://mitpress.mit.edu