UNIVERSES GALAXIES ETC UGSBOC Cecil J Phillips 1 SIG IPS ANNOUNCEMENT 8 WHY ARE THESE PEOPLE SMILING Mary Ann Harrison 9 CLA BANQUET ANNOUNCEMENT 12 CONTROVERSIAL BOOK ON ARTIFICIAL INTELLIGENCE J THE POLYHEDRAL WAR ooooooooooooooooooooooo I PLAIN ENGLISH 1- 21 A STORY WITH A MORAL 22 CHOOSE YE oooooo oo oo o oo oooooo 27 LETTER TO THE EDITOR oooooooooo o oooooo ooo oo oo oooooooooo 28 I t ' L 6-36 f ' IIiS B6etiMIN'I' e9N'F1 IN8 El9BIWQRB MATBRIAL n 'ho 9111N8A GIIGtII M QlIIM 111 Z lJDm t INa 61 8 10 11182 c 2 eclassified and Approved for Release by NSA on '10-'1 '1- 20'1 2 pursuant to E O '135 26 vl DR Case # 54778 DOClD 4009794 P E TOP SECRET Published Monthly by Pl Techniques and Standards for the Personnel of Operations MAY 1977 VOL IV NO 5 WILLIAM LUTWINIAK PUBLISHER BOARD OF EDITORS Editor in Chiefoooo 'Arthur J Salemme 5236s Collection 1 IC8955sJ Cryptanalysis 1 1 4902s Language o o o Emery W Tetrault 52 3 6 s Machine Support o 1 IrS303s Mathematics o o o Reed Dawson 3957s Special Research Vera R Filby 7119s Traffic Analysis Frederic O Mason Jr 4l42s Production Manager Harry Goff 4998s For individual subscriptions send name and organizational designator to CRYPTOLOG PI 'fOP SBCRBT P L 86- 3 6 86-36 4 e DOCID 4009794 CRI3T hen I was first asked by the SIGIPS program committee to talk about relationships among systems projects etc I almost laughed out loud because we had wrestled with a similar request by Admiral Phillips a couple of years ago What made Admiral Phillips' request impossible to answer was that I presume he wanted an official reply -- that is one which had been coordinated with the several thousand people -- almost everyone in the Agency -- who consider themselves systems designers and the like I am not sure I remember to whom Admiral Phillips' request was addressed but it was not addressed to C so we did not answer it All of this is by way of explaining that what I am going to say is pretty much my own opinion and that I hope you understand that there is not a single simple chart which will relate everything That is unless we take the simple approach of creating a new project -- standing for Universes Galaxies etc Everyone of these items I will discuss is mult ifaceted - - and many are perceiveddi fferently by different viewers Thus it is not surprising that one viewer can seea pair of projects as badly interfaced and another viewer can see them as having a satisfactory interface One more bit of philosophy before I talk about some of the specific projects I think that project management has a bad side as well as a good side On the good side it probably leads to better-defineciprojects bounded projects and hopefully to timely and correct completion of projects On the bad side it leads to introspective parochial views of the world and narrow problem approaches I be lieve project management needs a little more tempering -- with some input from disorganized geniuses Don't ask me how to distinguish geniuses from cranks I don't know the answer to this but I think we have a lot of both P L 86-36 EO 1 4 c May 77 CRYPTOLOG Pa ge I SBCRBT IIANf'iLE cIA e6li f ft' 811AtiNEbS 8NbY DOCID 4009794 SBEURRET Fig 2 P L 86-36 EO 1 4 c May 77 CRYPTOLOG SECRET Page 2 WA JQb lq 0 t9ShUN'F 8IlANN13LS 8NL'i L - DOC I --I Dh ------'4 0-0-9-1-9-4--- 86-36 1 4 SECRET May 77 CRYPTOLOG SECRET Page 3 IIANBhl3 YIlt 8SftllN'f' 8IIANNrJLS 8NI t 86-36 DOCID 4009794 1 4 c 8 BEURRB'F May 77 CRYPTOLOG Page 4 IIltN815J3 Vllt SSMINy SIIMlNJ3h8 8H15Y DOCID P L 86 36 EO 1 4 Ie 4009794 SBCRB'f May 77 CRYPTOLOG Page 5 SECRET Ih NBbB '1A eSMIn'f' SIiAPHil3bS eNbJPY DOCID 4009794 8BCRBi May 77 CRYPTOLOG Page 6 SHURET IIIiPIQJ S VI' 8SMHoI''t' eIlAlm hS hLi DOCID 4009794 P L 86 36ED 1 4 Ie 8 CR 'I' 7 May 77 CRYPTOLOG Page 7 SECRE'I' fh NBbB fA EURleMHi EURlIIldHlBbS ePibY DOCID 4009794 SBCRB'f 5EeRE'l' 1IIIIm U The preceding aiticle is the text of a presentation given b Mr Phillips to members of the special Interest Gl'OUp on Info1'mation Processing Systems fSIG IPS in August 1978 SIG IPS is an organization under the auspices of the NSA Computer and Info1'mation Sciences I eee Insti tute CISI which addresses prob lems and technical developments in the fields of data storage data processing and data retrieval Mr Phillips' presentation was one of a series of monthly programs which SIG IPS sponsors for the benefit of all interested NSA emp'loyees Additional pl'Ograms include seminars and technical films Persons wishing to join SIG IPS 01' to suggest topics tor r o grams are invited to contact IG IPS Chairman on x5504s UNCLASSIFIED May 77 CRYPTOLOG Page 8 P L 86-36 EO 1 4 c SECRE'f P L IIMlBbE t e8MH I't' eIlAN'H13M 86-36 OI Li DOCID 4009794 SECRET r XJ d ' WHY I ARE THESE PEOPLE SMILING Mary Ann Harrison A64 ctuallY I have no idea But now that I have your attention I hope you will continue reading to find out why certain other people the Agency's foreign-language transcribers -- are definitely not smiling ance of transcription there is widespread lack of understanding and for that matter concern about what a transcriber does and what it takes to be a transcriber Many NSA managers and supervisors especially those who themselves have no foreign-language training do not realize how difficult transcription is In their ignorance they dismiss the work as a purely mechanical function and transcribers as nearautomatons Despite his l position as the linchpin in the SIGINT process the transcriber has for many years been the stepchild of the intelligenceprocessing family While ost professionals in the SIGINT business acknowledge the import- Unfortunately this attitude toward transcription is not confined to those in management and supervisory positions Newly hired linguists are indoctrinated with this negative evaluation of transcription even before they are on the job It is not unusual to hear recently arrived college-trained linguists say t ey will quit rather than go into transcrip 10n even before they know what transcription s The anti-transcription virus has apparently spread to recruiting and personnel offices II shall use the masculine forms throughout in order to avoid the cumbersome he or she constructions and also because the majority of transcribers are currently at least men Every transcription shop has its own horror stories to illustrate the prevailing ignorance indifference and outright hostility to transcribers and transcription I will not burden you with many of them but I believe a few excerpts will illustrate my point EO 1 4 c P L 86-36 May 77 CRYPTOLOG Page 9 II 1 JQbE ' 'IA C8P HPJ'T eIL tHJEbS 8tH Y DOCID 4009 'C'l9 c P L 56-36 ISCKIST What other gro p of N5A professionals has heard themsel ves described by a senior management official as' 'pl umbers The connotation was unmistakable tr anscribers are the daylaborers of the intel ligence world Analystreporters and others with whom the Russian tran scribers work on a daily basis have asked Where do you keep yo r transcribers Or have issued instructions to take these tapes and run 'em through your transcribing machines There is a real and apparently widespread belief that transcribers do nothing but thread tape through a machine turn it on and collect page-print as it pours out the other end Many of those who understand that areal live human being is involved in the transcription process still fail to appreciate the difficulties of the job These persons I assessment of transcription is that any idiot with a dictionary can handle this stuff So much for the transcription myth Now please allow me to present a thumbnail sketch of who a transcriher i wh r he rlnp nrl hnw May 77 CRYPTOLOG Page 10 SECRET II taJQbE HI' GQPlINT GIl ' l'JNEbe Ql'IbV t DOCID 4009794 SEeRE May 77 CRYPTOLOG Page 11 SECRET HA1J8hE VIz eeflflN'f' EURlh UNBbS aUt i DOCID 4cr09794 U'lsl r I r c P L SECRE - - -------- - 86-36 - SE6RE'f 669 this date The date '12 18 3 17 16 - 13 Men 13 Zotz in the Maya calendar is an important one for those interested in the ancient Maya civilization On that date Charles Lacombe will make the question Does Anyone Here Speak Ancient Mayan the theme of his speech at the annual banquet of the Crypto-Linguistic Association Last July during a visi t here from his home base in Miami Mr Lacombe a former NSAer spoke to a joint meeting of CLA' s special interest groups on lexicography SIGLEX and translation SIGTRAN on computer techniques for decryption of Maya climatic records To make your banquet reservation call Steve Bladey P16 x5236s Oh incidentally for those who are a bit rusty on Maya dates the banquet will be on Thursday 23 JUNE May 77 CRYPTOLOG Page 12 SBEURRIST UNCLASSIFIED DOCIO 4009794 UNCLASSIFIED P L 86-36 Another Controversial Book on ARTIFICIAL INtELLIGENCE IEVIEWID IY PI3 n 1972 Hubert Dreyfus published a book entitled t Computep8 Can't Do Dreyfus 1972 This book created somewhat of a stir in computer circles a number of provocative challenges were made and consider able healthy and constructive self-examination was stimulated in the minds of many workers in the field of Artificial Intelligence AI It seemed that Dreyfus had raised so many basic questions that his book deserved a careful reading by anyone interested in the future of computer science even if only to strengthen and clarify the grounds for disagreeing with him see my review Machine Intelligence Promise or Delusion CRYPTOLOG July 1975 Now we find ourselves challenged again and this time on even more fundamental and extensive grounds in a new book Computer Powep and HlQ1IfZrl Reason by Joseph Weizenbaum of MIT published by W H Freeman and' Co San Francisco 1976 Dreyfus is a professional philosopher and his expertise lies entirely outside of computer science thus he was assai 1ing the bastions of AI research from an external position In Weizenbaum by contrast we see an insider -- a very highly respected and creative worker in the field of AI research -breaching the walls of the citadel from within Weizenbaum's accomplishments include the development of a useful list-processing computer language called SLIP Symmetric List Processor and various other work centering around the problem of making computers capable 0 language understanding and other intelligent human activities cf Weizenbaum 1963 1967a 1967b Along with his technical work Weizeribaum has also consistently carried forward an interest in and concern for the social implications of computer technology Perhaps his best-known AI wqrk was the development of a natural language analysis program called ELIZA which could carryon a dialog on some preselected topic with a human conversational partner ELIZA created considerable interest and excitement and also enjoyed substantial popularity as a demonstration vehicle in computer science centers around the United States in a version called DOCTOR which was prograouned to parody the interview of a Rogerian nondirective psychotherapist with a new patient In his introduction Weizenbaum sets the stage for the reader by describing his own frame of reference and stating his reasons for writing the book As a practicing computer scientist and a teacher of computer scienc in the great temple of technology which is MIT Weizenbaum became acutely aware of his students' questions and doubts about the value of what they were learning and doing On the one hand they were bending all their efforts toward learning their craft and many of them were seeking only to increase their dogmatic dependence on and polarization around science and technology On the other hand many were aware May 77 CRYPTOLOG Page 13 DOCID 4009794 UNCLASSIFIED that they were not learning anything that would help them to choose the right questions to ask or to evaluate the answers they got Some were rebelling openly at working on projects that appear to address themselves neither to answering interesting questions of fact nor to solving problems in theory p 11 In addition to the questions raised in the cfassroom Weizenbaum was confronted with a more abrupt and compelling philosophical shock as a result of his work with ELIZA He saw practicing psychiatrists hailing what was really no more than a technical trick of parodying one particular interviewing method as a breakthrough in automated psychotherapy He was further dismayed to see how quickly and easily people of all walks of life and degrees of education became emotionally involved with ELIZA playing DOCTOR accepting it as a therapist and personifying it to the extent of asking Weizenbaum to leave the room so that they could converse with the DOCTOR in private A third unpleasant surprise emerged when he became aware of a widespread belief that ELIZA represented a general solution to the problem of computer understanding of natural language These disproportionate overevaluations of his successful but modestly-framed experiment made him realize that most people do not understand anything about computers and are ready and eager to believe that these magi cal machines and those who program them can do anything under the sun Weizenbaum tells us that he felt himself thus suddenly faced with a terrifying responsibility that seem to compel him to step back and reconsider in a new light what he had been doing His book is the result of a 2-year leave of absence from teaching one year spent at the Stanford Center for Advanced Study in the Behaviorial Sciences and the other year at Harvard University both on research grants His work at these institutions benefited greatly by contacts with many other workers in a variety of fields some of them well-known authorities and his manuscript was read and commented upon by these new colleagues How have we reached the point where practicing psychiatrists can so overvalue a computer program as a therapeutic breakthrough What historical process has brought the man in the street to the place where he can unquestioningly accept a computer program as an appropriate receptacle for his deepest emotional and personal confidences Weizenbaum poses these questions and seeks to answer them by tracing the transformations wrought upon human thought and experience through the growth of new tools and techniques that laid the foundations for modern empirical science He seeks to show that this history and the particular paths it has taken have brought us to a scientific outlook that seems to have produced a mechanistic conception of man We have made ourselves and May 77 all our institutions over into the image of the computer A practicing psychiatrist under this mechanistic spell sees nothing amiss in a picture of himself not as an engage human being acting as a healer but as an lnformatlon processor following rules p 6 Weizenbaum's trip back into the history of Western thought begins with the clock -- man's first truly autonomous machine capable of running for long periods without human guidance or intervention He follows Mumford 1963 in stressing the importance of the clock and its use by medieval monks to time their devotions in the monasteries With Mumford Weizenbaum sees the widespread use of clocks and measuring devices producing pointer readings and numb r as the cause of a crucial change in man I s perception of time and consequently of space The concept of time as disassociated from particular events made possible the creation of ' an independent world of mathematically measurable sequences the special world of sc ience' Mumford's words quoted by Wei zenbaum p 23 The clock as man's first autonomous machine foreshadowed the programmable computer both once started continue operating on the basis of an internalized model of some aspect of the real world Clocks are fun damentallymodels of the planetary system p 23 The growth of our dependence on machines of various kinds and the changes that brought the modern scientific world view into wide acceptance resulted Weizenbaum continues in a wholesale rejection of direct experience Gradually at first then ever more rapidly and it is fair to say ever more compulsively experiences of reality had to be representable as numbers in order to appear legitimate in the eyes of the common wisdom p 25 The value of everything and especially of human labor became quantified as money These developments set the stage for such devout and impassioned statements of the ascetic faith of science as that of Pearson dating from 1892 which Weizenbaum quotes 'The scientific man has above all things to strive at self-elimination in his judgements ' p 25 This statement as Weizenbaum points out urges man to make himself into a disembodied intelligence a sort of dispassionate dehumanized machine The computer arrived on the scene after the scientific transformation of human thought was essentially complete Weizenbaum describes the first widespread practical use of computers automating the great tab rooms of business and industry in performing their accounting operations In this use the computer took over essentially unchanged the work of punched-card accounting machines that sorted summarized and printed office records Then businesses and government agencies found that the very techniques they had developed to study and model their own operations often motivated by the acquisition of their first computer could be CRYPTO OG Page 14 UNCLASSIFIED DOCID 4009794 UNCLASSIFIED programmed and carried out by means of the new machines The crucial transition from the business computer as a mere substitute for work-horse tab machines to its present status as a versatile information en ine began when the power of the computer was proj ected onto the framework already established by operations research and systems analysis pp 33-34 It has become a truism among spokesmen for technology that the computer came along just in time to save our society from breaking down' under the strain of population growth and exploding complexity Wei zenbaum while agreeing that our institutions could not now exist in their present form without the computer provides a different slant on the matter He shows how the computer was seized upon by business and government as a God-given means of maintaining the status quo and sidestepping the creative reshaping of our institutions that might have been called forth by the crises in complexity that threatened after World War II The bottlenecks and breakdowns that motivated us to augment or replace the low-internalspeed human organizations with computers might in some other historical situation have been an incentive for modifying the task to be accomplished perhaps doing away with it altogether or for restructuring the human organizations whose inherent limitations were after all seen as the root of the trouble p 30 Weizenbaum condemns the choice that wa6 made and the use that was made of computers as a result of that choice in words recalling some of Dreyfus' statements Of the many paths to social innovation the invention of the computer opened to man the most fateful was to make it possible for him to eschew all deliberate thought of substantive change That was the option man chose to exercise pp 31-32 The computer has now become truly indispensable to the survival of our society in the present speaifia form in whiah the aomputer itseZf has heZped to freeze and fossilize one way of doing things Weizenbaum devotes several very readable and interesting chapters to an explanation of what computers are how they work and whence they gain their unique and awesome power to change our world His excellent clear discussion of the Turing machine the nature of programs and programming and other computer concepts are in my opinion worthy of unreserved recommendation to any reader who wants a good insight into the matter Whatever disagreements one may have with his value judgments Weizenbaum must certainly be a superlative teacher of computer science The computer Weizenbaum continues perhaps more than any other tool has altered Man's concept of his own identity mind and experience An all-pervasive computer metaphor has taken hold of the minds of scientists many humanists and the man in the street According to this view man and computer are merely two subspecies of the genus information processor Weizenbaum singles out for special attack the Simon and Newell General Problem Solver GPS and its means-end analysis of human problem-solving behavior This theory sought to found a new information processing psychology based on a hierarchical set of modular actions and tests which described all goal-directed behavior in terms of reducing differences between the object in hand and a desired object Simon and Newell 1958 The broad and ambitious general claims made for the GPS theory by its authors are roundly condemned by Weizenbaum who dismisses it as a limited metaphor and not a scientific theory at all in any proper sense Its worst consequence in his eyes is the production of an image of Man as a GPS-like machine This impoverished metaphor says Weizenbaum has made it possible for a psychiatrist to regard his patient as an object different from the desired object The therapist's task is to detect the difference using difference-detecting operators and then to reduce it using difference-reducing operators and so on That is his 'problem' And that is how far the computer metaphor has brought some of us p 181 Weizenbaumpoints out a number of other consequences of the overenthusiastic and unbalanced application of computer technology While new tools can open up new horizons for people he warns us that they can also have the effect of closing off certain possibilities of social action and distancing us from some domains of data and experience that were previously accessible A computing system that permits the 'asking of only certain kinds of questions that accepts only certain kinds of 'data ' and that cannot even in principle be understood by those who rely on it has effectively closed many doors that were open before it was installed p 38 Another important adverse consequence of the modern scientific world view and especially of its latest manifestation the computer is an abdication of individual human autonomy and responsibility and a new tyranny of technical expertise Weizenbaum condemns a number of recent theories and research efforts among them GPS Forrester's Limits to Growth model and B F Skinner's psychological theories accusing them of distorting and abusing language advertising easy scientifically-endorsed panaceas for every ill and exploiting the myth of expertise The language is mystifying All contact with concrete situations is abstracted away Then only graphs data sets printouts are left And only 'we ' the experts can understand them The expert will take care of everything even of the problems he himself creates pp 253-254 Weizenbaum presents a provocative viewpoint on science itself that most sacred of sacred cows He accuses it of having become in our present overemphasis and abuse an addic- May 77 CRYPTOLOG Page 15 UNCLASSIFIED DOCID 4009794 UNCLASSIFIED tive drug and even a slow-acting poison in spite of the admitted benefits we have obtained from some of its accomplishments Science directly contrary to the generally accepted belief in its solid foundation in absolute objectivity and the validated work of the past can only be anchored ultimatel on the shifting sand of fallible human judgment conjecture and intuition p 15 The myth of scientific objectivity would have us believe than any clearly demonstrated counter-instance is sufficient to force revision of the most cherished theory On the contrary WeizenbaUm shows scientists routinely explain away counter-instances as erroneous or inconsequential and cling to their pet theories in the hope that they will be rescued by some later discovery The man in the street surely believes scientific facts to be as well-established as well-proven as his own existence His certitude is an illusion p 15 Drawing a trenchant and amusing parallel between the compulsive gambler the compulsive programmer and the mad scientist Weizenbaum drives home some telling points All three share these characteristics in common they are certain they will succeed regardless of any number of setbacks they have an invincible faith in their own cleverness and they are convinced that all of life can be reduced to the terms of their system whether it be gambling programming or empirical science All three disciplines are revealed by Weizenbaum as magical and self-validating belief systems capable of being unrestrainedly abused by people with an overriding passion for absolute certainty and control over artificial worlds of their own making The compulsive programmer is merely the proverbial mad scientist who has been given a theater the computer in which he can and does play out his fantasies p 126 In an interesting discussion of theories and models Weizenbaum contrasts the theory-based application of computers in such successful AI projects as CSYMA and DENDRAL to the ad hoc patchwork approach in most other areas where no strong theoretical base exists or has been applied DENDRAL at Stanford University produces descriptions of those molecular structures that can explain a given record from a mass spectrograph Its competence is as good as or better than that of the human postdoctoral chemists who have traditionally performed this task Buchanan Sutherland and Feigenbaum 1969 MACSYMA at MIT performs symbolic mathematical manipulations also usually carried out by highly-skilled specialists Martin and Fateman 1971 Such theory-based programs Weizenbaum suggests enjoy the enormously important advantage that when they misbehave their human monitors can detect that their performance does not correspond to the dictates of their theory and can diagnose the' reason for the failure from the theory p 232 Most existing programs in use by business industry and government can be based on no such theory they are Weizenbaum maintains patched and hacked together from hit-and-miss strategies that seemed to work for this or that case Such programs are typically written by large loose teams of programmers many of whom have left or been reassigned before the program goes into operation Thus there is no one around who really knows how the program works in the absence of any theoretical base no one can even tell whether it is working properly in all cases let alone correct it when it fails Weizenbaum justly devotes considerable attention to Terry Winograd's language-understanding research This work undoubtedly constitutes one of the most successful and creative recent AI accomplishments and has been hailed by many as a major breakthrough I strongly recommend Winograd's report 1970 to any reader interested in gaining an idea of what can be done by computers in the analysis of natural language it is enjoyable reading and represents a very fine piece of research Weizenbaum does not see it however as a theory-based accomplishment comparable to MACSYMA or DENDRAL He praises it as an important achievement which shows how a specific view of certain aspects of language can really be filled in with enough detail to provide a working model p 195 Winograd's heuristics however according to Weizenbaum express no interesting general principles p 197 Winograd's effort shares with GPS and many other AI projects the fault of appearing to present general theories when they really provide only virtually empty heuristic slogans They 'seek to verify these theories by constructing models that do perform some tasks but in a way that fails to give insight into general principles p 196 Weizenbaum bases his deeper philosophical objections to such AI research and to the abuse of computers and of technology in general on two main grounds His first argument is an attempt to show that much of human thought is not amenable to formal expression His second basic position concerns the ways in which man must differ in essence from any machine In much of what he says on these topics Weizenbaum comes close to many of Dreyfus' points but with a rather different emphasis Human thought and behavior cannot ever be entirely expressed in terms of the effective procedures which can be carried out by a Universal Turing Machine The reasons he presents for this view can be summarized as follows o Much human thought action and speech is context-dependent It matters quite a lot to us which values of x and y are plugged into an expression The human participant in a conversation or other social act brings to it a frame of reference that establishes what the talk or action is about Weizenbaum maintains that we can never get a sufficiently inclusive knowledge base into a computer to permit it to do this for any significant task or context Continued on p 23 May 77 CRYPTOLOG Page 16 UNCLASSIFIED DOClD 40093 9 4- - - - - -- -- --- -- UNCLASSIFIED From 0 to the undrawable THE POLYHEDRAL WAR lpI2uu_u_uuuu ILLUSTRAfED IY uu I J2 1l- ome time ago a model of a polyhedron Little did I know that my innocent boast appeared in the PI office For those of you would begin the now-famous Polyhedral War The who have forgotten -- or never had -- threesame one-upsmanship th lt'led me to build a betdimensional geometry a polyhedron is the ter model would alsQ -iead another young mathespace-filling analog of the area-filling polymatician the ill strator of this article gon A square is a two-dimensional polygon 'to do the same Yet consider a cube is its three-dimensional analog ' th e-- 's-s-u-e--a b-o-u twhich the War was being waged a polyhedron In their idealized form the beauty of models of abstract geometric enpolyhedra are examples of some of the m st tities This may serve to show the only real beautiful shapes and CaR often be seen ncor- distinction between the outl ok of mathematiporated into the works of contemporary artists cians and that of others Often the mathemati- Yet the model of the polyhedron in the PI cian is concerned with more abstract things office a flexible dodecahedron 12-faced than the general person yet the manifestation figure in basic black was at the opposite of that concern takes on very familiar practiend of the beauty scale see Fig 1 It cal forms e g a contest This contest the was in my opinion one of the most pitiful War is a good example of this dichotomy beexamples of a polyhedral model I had ever seen tween the planes of thought and of action The To add insult to injury in addition to being War was fought at two levels -- an abstract unaesthetic it was constructed by using a most level with choices of polyhedra or polyhedral inefficient method Both of these consideracombinations and a practical level with contions led me to state that I with my backstruction methods materials colors and glue ground in geometry a d my interest in modelMichael chose models of interpenetrating building could easily build a better polyh dra model the only criterion of better be ng ts polyhedra as the most beautiful These figures known since the time of Gauss and Euler often aesthetic quality as judged by an unbiased obdepict graphically the interrelationships among server the five most basic polyhedra the Platonic solids These solids are o tetrahedron tetra four hedl'on face o cube figure with six square sides o octahedron octa eight o dodecahedron dodeca twelve and o icosahedron iaosa twenty Fig 1 Pitiful polyhedron They exhibit a surprlslng degree of interdependence The cube and the octahedron for example are duals Le mirror images in a certain sense More specifically the cube's six faces eight vertices and 12 edges correspond directly to the octahedron's six vertices eight faces and 12 edges Indeed if you connect the centers of the six square faces of a cube you obtain an octahedron and vice versa with the eight triangular faces of the octahedron as can be seen from Figs 2a and 2b These and May 77 CRYPTOLOG Page 17 UNCLASSIFIED P L 8 6 - 3 6 DOCIO 4009794 UNCLASSIFIED a Fig 3a One tetrahedron in a dodecahedron Fig 3b Five tetrahedra visible edges only in an imaginary dodecahedron b Fig 2 Duality of cube and octahedron a - Octahedron in cube b - Cube in octahedron other unexpected and unusual properties have fascinated mathematicians and mystics alike even so far as to have inspired various schools of cosmology into varied and numerous misguided interpretations of the mathematics involved With this wealth of history behind him Michael decided to build a model of five intersecting tetrahedra intersecting in such a way as to out ine the largest of tne Platonic solids 1 the dodecahedron This polyhedral configuration is pictured in Fig 3 After choos ng the basic strategy for using these models of l nterpenetrating polyhedra Michael worked out the tactics of construction He chose to use the traditional method of cuttlng and pasting faces which are especially constructed for the particular model This method requires extreme exactitude in the preparationof the faces and great patience in the assembly and is suited only for permanent-display models which represent polyhedra whose geometry has been completely de cribed lWhile the dodecahedron has fewer faces than the icosahedron it is larger in terms of volume The ancient Greeks knew this and made balls for their sporting events with 12 pentagonal faces sewn together The family of polyhedra I chose however is known as the toroids A toroid is a polyhedron out of which a polyhedral hole has been cut As an example consider Fig 4 a square orthobicupola with a cuboctahedron hole cut out of its inside These types of polyhedral combinations show the relationsh ps between the altitudes and angles of the varl OU polyhedra as opposed to the interpenetration figures which tend to reveal the symmetry relationships Also unlike the interpenetratl on figures toroids are relatively new most having been discovered since 1970 As an example of one of the most aesthetic toroids I chose a three-holed truncated octahedron Since this toroid is extremely difficult to visualize from a drawing Fig 5 shows how l t is constructed as well as the final result May 77 CRYPTOLOG Page 18 UNCLASSIFIED UNCLASSIFIED a Fig 4 The construction of a simple toroid a - Components b -- Final result Toroids as a group represent a real challenge to the model-builder The traditional cut-and-paste method fails in two important areas The construction of a many-holed toroid as that depicted in Fig 5 requires an alm st unattainable degree of dexterity and patlence The experience of using this method is so discouraging that some mathematicians have suggested that the lack of an appropri te model-building technique may account for the fact that toroids have been unknown for so long Also the cut-and-paste method is totally unsuited for any kind of exploring Often it would be nice to build a quick model to assist one in working out the geometry involved in a polyhedron in a manner comparable to the way in which proofs in high-school geometry are solved with the aid of a rough sketch One cannot quickly test out an idea for a toroid with the cut-and-paste method In the first place the inspiration for the idea would probably die before a model could be constructed' with this laborious method and in the second place this method requires that the geometry of the polyhedron be entirely determined before bUilding the model But the model was being built to assist in the working out of the geom May 77 Fig S a - Exploded view of the hole b - Shell into which components of hole are fitted c - Final result -- a three-holed toroid These weaknesses have led to the so-called cardboard-and-rubberband method In this method regular polygons are constructed in great numbers without having any specific polyhedron in mind These polygons are appended with flaps on each side The flaps are bent to be at right angles to the polygon and notched' at each vertex Two such polygons can be joined in the process of building a polyhedral model by a rubberband surrounding the flaps which is caught securely in the notches While it is true that the resulting models have flaps and rubberbands on the outside which tend somewhat to highlight the edges and vertices of the polyhedron they are mainly an aid in construction and are not a part of the idealized geometric entity Fig 6a shows a polyhedral model complete with these flaps and rubberbands For explorations this method yields an unexpected bonus Models made in CRYPTOLOG Page 19 UNCLASSIFIED DOClD 4009794 UNCLASSIFIED this way may be quickly disassembled and the pieces can then be used in a mild form of cannibalism for new models The effort expended to construct the original polygonal pieces then becomes an investment as the p1eces can be re-used While the ease ot use and the economy of effort of the cardboard-andrubberband method cannot be overemphasized it must be remarked that the final model lacks the professional appearance of the cut-and-paste models and also that unfortunately rubberbands do not last forever After a year or so one can be horrified to find his cherished model as a tangled jumble of polygons and decayed rubberbands The cardboard-and-rubberband method of construction along with a lengthy and delightful exploration of the world of toroids is fully described in Stewart's Adventures Among the Toroids The toroid in Figs 6a and 6b a pentagonal orthobicupola wi th a pentagonal prism as a hole reminds us that the War is fought on two levels and not merely on one For in reality Fig 6 Fig 6a Cardboard-and-rubberband model represents nothing at all While examination of the figure or even the model itself reveals nothing suspicious a detailed analysis of the geometry of the model shows us that such a oroid is nonexistent Assuming an edge lengtr of one for all polygons the altitude from the parallel pentagonal faces of the bicupola would have to be precisely one for the prism to fit exactly inside After a small amount of computation the altitude of the bicupola is found to be h where R 1 2 jR 2 5 _ 2 This yields 1 05 h 1 06 Thus the idealized penta onal prism could ot possibly fit inside the idealized pentagonal icupola The model appears to fi t together beause there is just enough stretch in the ruberbands so that the distortion is not noticed hen the model of the bicupola is compressed he War requires not only attractive models ut also that the models represent an actual geometric form Cardboard and rubberbands do not a polyhedron make While the tide of the War was decidely in my avor with the submission of the three-holed runcated octahedron built by using the cardoard-and-rubberband method Michael after much ecretive delay and deception countered with two flawlessly constructed interpenetration figures One was the nonsuperimposable mirror image of the five intersecting tetrahedra and the other Figs 7a and 7b was a model of five intersecting cubes also intersecting in such a way as to outline the dodecahedron At this point however because of time and space limitations the PI office was beginning y Fig 6b Top view of idealized cardboardand-rubberband model rubberbands and flaps removed Fig 7a One cube solid lines with positioning of next cube dotted lines May 77 CRYPTOLOG Page 20 UNCLASSIFIED - -- --------- - --- - _ DOCID 4009794 UNCLASSIFIED of 260 faces was declared unsurpassable by the unbiased judge of the contest and can now be viewed in the PI office Room 3W090 Since a drawirig can hardly do justice to a model of such complexIty and-beauty none is even attemp here Fig 7b Perhaps the average person thinks that the mathematician gets his fun out of standing hipdeep in computer output as he mutters absentmindedly about some obscure calculation or by filling in blackboards with theoretical equations so much so that he is completely unable to enj oy ordinary pleasures Actually mathematicians are pretty much 1ike everyone else and can appreciate a lot of other things too including beauty wherever they see it And one of the biggest joys of the mathematician is to use mathematical principles to construct a beautiful object that everyone can enjoy in his or her own way -- the mathematician because it is a truncated dodecahedron logically derived from one of the Platonic solids and the average person just because it looks absolutely beautiful Five cubes in a dodecahedron visible edges only to get rather crowded a truce was declared and the War officially ended by the joint submission 2 of a model of a toroid constructed by the cut-and-paste method The torQid chosen was Stewart's Gem of the Expedition a truncated dodecahedron with 11 holes constructed of pentagonal cupolas and antiprisms This gargantuan effort consisting I I provided invaluable assistance in the construction of this model AU the mode lA ill t L6 tJtate d in thM aJL ticte piU6 the undJLawa bie Gem 06 the Ex pe ciLUon wLU be hown in the Hea dquaJLteM BGBPLi t Ung ciJ6pi a y c Me -i n the pM a gewa y towtVtd the Ope Jt i tio uUcUng J 6Mm 16 tM ough 31 Ma y s PLAIN ENGLISH I e22 Repr inted from C-LINERS Final issue OOk Mary See Bobby write Bobby writes bad He writes long words Bad bad Bobby See me write I write good I write little words Good good me And I am a college graduate Daddy paid lots and lots of money so could speak English good I was an En 11sh major Now I work I work at NSA Now I use APE Agency Plain English Now I am good I even got a letter of appre o appri P L 86-36 apre o I forgot how to spelliJ Rubbish Understangthatl am not against plain communication straightforward communication where each word must be justified must add a necessary and vital ingredient to the communication But our nation's universities are horrified at what they have come to call college-level illiteracy This year in the California state colleges nearly 50 percent of the incoming freshman class was required to take mandatory noncredit remedial English Yet at NSA we are encouraged to all but abandon what we used to call the English language We have actually been told not to worry too much about formal sentence structure We have been cri ticized for using a dictionary and thesaurus But did it ever occur to anyone that wordy passive confusing memos might be a weapon or more correctly a shield If we want all Agency personnel to speak and write plain English perhaps we should first teach Agency personnel English If we want Agency management to write concise active decisive memos perhaps we should first teach Agency management to be concise active and decisive Let us attack the problem not just hide the symptom May 77 CRYPTOLOG Page 21 UNCLASSIFIED DOCID 4009794 UNCLASSIFIED A STORY WITH A ORA Ma ny 6p uu te d fucU6 6' 'oM have be en hel d by the TA Tvun i lWlogy Pa nel d uM ng the 4 ye aJlA 6pe ttt 'J pda ti n g the 1958 Comb ned G lo 6 6aJttj TJLaU c An alytic TeJrmin o e 081J On e di 6c lL6 qn o duJUng whi chthe Chiii JlpeJL60n 6iUd Let' 6 Ite v w the bM' 'C teJu1l one n n a l time - - tha t 6 6OIL the ump teenth time - - ItU lLUe d n the 6oUow i n g on To all the members Panel In the beginning God created a TERMINAL For a while this terminal was happy He had a transmitter and a receiver He played with the transmitter sending little signals throughout the ionosphere But although he could send he never received any answers So the little terminal became unhappy After much thought God created another terminal and He called the two terminals a LtNK Because He was a kind God He assigned them a single frequency to use for communication and He called this SIMPLEX WORKING 6 tolty N G chatted happily together One day as happens in all good stories one of these terminals decided that he wanted to talk to two of his friends simultaneously and he needed two frequencies to do so Control agreed and called it DUALING Now JealOusy reared its ugly head Another terminal decided he also wanted two frequencies but he wanted to transmit different information simultaneously Control was shook But he said okay and he called it DOUBLE CHANNEL OPERATION Two other little terminals who had been Time passed and as is wont to happen in the best of families the two little terminals each watching all this byplay decided they too wanted to claim the frequency for its very own shOUld get into the act They wanted to simultaneously transmit and receive messages in both So God in His infinite wisdom gave each terdirections between themselves each using his minal its own frequency and He called this one transmitter and his one receiver Control COMPLEX WORKING began to worry thinking This is getting out More time passed and almost nine months of hand But he said okay and he called it later to the day give or take a few days a DUPLEX OPERATION new terminal appeared God decided He would But he added just in case it should arise call these terminals a GROUP the normal practice of transmitting and receivNow He knew from past experience that ing messages between two terminals in either trouble could arise so He made the first direction alternately would be called HALF terminal CONTROL And He told Control You DUPLEX OPERATION And he went home and are the boss of this group and you tell 'em threw up how it' s gonna be Then God came home And He left on vacation He looked around He could not believe what Now Control decided that there was only one He saw way to make everybody happy so he gave each He called Control and asked What happened terminal a different transmitting frequency to use to contact the other terminals and he Control replied The devil made me do it called it COMPLEX SENl ING God said I thought when I told you you're Additionally because he had some smarts of his own Control gave each terminal a different the boss you would have sense enough to use the KISS principle That's Keep It Simple Stupidl receiving frequency and all terminals when sending to a particular terminal used the freThe moral of the story is quency allotted to it And he called it Terminals and panels should not mess with COMPLEX RECEIVING God's creations Still more time passed and the group grew $ More terminals appeared and these terminals all May 77 CRYPTOLOG Page 22 UNCLASSIFIED P L 86-36 DOCID 4009794 UNCLASSIFIED Artificial Intelligence Continued from p 16 o MUch human thought is essentially beyond Weizenbaum refers to the reformaZization cent work of Ornstein 1972 and the psychological discoveries concerning differences of function in the left and right hemispheres of the human brain According to this still somewhat new and controversial finding the left hemisphere specializes in sequential logical and verbal thought while the right specializes in holistic global intuitive concepts and movement in space We have thus at least two major modes of thought the left-brained mode produces structures like computer programs and formal descriptions but the right-brained mode so far completely escapes our ability to pin it down in formal terms We must use our left-brained methods to think rationally and science singles these out as the sole avenues of thought for its purposes Unfortunately for science and for AI research human problemsolving and communication require that both halves of the brain work together in close cooperation Dreyfus' distinction between digital and nondigi tal modes of thought and his mention of fringe consciousness insight and perspicuous grouping as nondigital features of human thinking appear to be closely related concepts o FormaZ systems and empiriaal observations are essentially inaomplete Weizenbaum alludes to Heisenberg's Uncertainty Principle as undermining the basis of emprical observation on which science is founded further Gl ldel' s Proof-exposed 'the shakiness of the foundations of mathematics and logic itself and proved that they must necessarily be forever incomplete p 221 Thus there must always be things that we can never measure accurately or describe COmpletely o Human beings are errbodied Here Weizenbaum echoes in' almost the same words one of Dreyfus' most important and interesting points It is not obvious that all human knOWledge is encodable in 'information structures however complex there are some things humans know by virtue of having a human body No organism that does not have a human body can know these things in the same way humans know them Every symbolic representation of them must lose some informatio that is essential for some human purposes pp 208-209 o Human beings are Boaialized by other humans The human infant is born prematurely that is in a state of utter helplessness infant will die if he is fed and cleaned but not from the very beginning of his life fondled and caressed -- if in other words he is not treated as a human being by other human beings Thus begins the individual human's imaginative reconstruction of the world And this world is the repository of his subjectivity the stimulator of his consciousness and ultimately the con- structor of the apparently external forces he is to confront all his life pp 210-211 By virtue of facing and solving human problems man's thought is determined in ways forever and essentially alien to the procedures and represen tations of a computer Man is not a machine Computers and men are not species of the same genus p 203 We are faced as Weizenbaum vividly demonstrates with a basic epistemological and moral dilemma Science provides us with tools in the form of descriptions algorithms and computer programs we have employed these tools to gain an unprecedented degree of apparent control over the physical world Science also provides us with a seductive and all-encompassing way of perceiving the world in terms of numbers rules and formalisms We are armed with a plethora of instruments and devices but we have cut out from under ourselves the basis of making value judgments or for choosing whither we shall direct our armament of techniques Science rules out by def1nition all the vast domain of reality that is subjective that cannot be counted measured or proved and where the bases of human choice truly reside Weizenbaum makes repeated use of a metaphor drawn from an old joke involving a drunk who has lost his keys on a dark street at night but persists in looking for them only under the street lamp because the light is so much better over here This is the way science and indeed all of man's left-brained undertakings must proceed The trouble comes when the scientist or programmer deliberately and consciously simplifies and thereby impoverishes and distorts reality and then proceeds to accept that distortion as a complete and final picture claiming to have produced a theory about the world when he has only pulled off a limited technical trick Indeed what is sought can be found only where there is illumination Sometimes one even finds a new source of light in the circle within which one is searching Two things matter the size of the circle of light that is the universe of one's inquiry and the spirit of one's inquiry The latter must include an acute awareness that there is an outer darkness and that there are sources of illumination of which one as yet knows very little p 127 Since our scientific lamppost has no light to throw on any matter concerned with human values or purposes we are in the position of the compulsive programmer who has forgotten if he ever cared what his program was supposed to do in the first place He has only technique not knowledge He has nothing he can analyze or synthesize in short he has nothing to form theories about His skill is therefore aimless even disembodied It is simply not connected with anything other than the instrument on which it may be exercised p 118 Enthroned in the midst of our artificial world May 77 CRYPTOLOG Page 23 UNCLASSIFIED DOClD 4009794 UNCLASSIFIED and all our expensive gadgets Iwe can count but we are rapidly forgetting how to say what is worth counting and whyu p 16 Weizenbaum anticipates some of the objections that will be raised against his positions The technologist will dismiss his arguments as merely philosphical and condemn them as polemical and subjective Such object are essentially just another way of saying that Weizenbaum's points belong to the world beyond the edges of that circle of light cast by the scientist's lamppost He makes the only answer he can repeatedly reminding the technologist that his keys are not where he is looking however good the light may be there A common defense of our total reliance on science and technology maintains that the only alternative is chaos destruction and mindless irrationality Weizenbaum counters this argument in a forthright statement In fact I am arguing for rationality But I argue that rationality may not be separated from intuition and feeling I argue for a rational use of science and technology I urge the introduction of ethical thought into science planning pp 255-256 The inevitable refusal of those on the opposite side of the debate to accept as legitimate any of the grounds on which he bases his arguments confronts Weizenbaum with an impossible dilemma Anyone who wishes to persuade his colleagues to cooperate in imposing some limits on their research o risks being excommunicated as a sort of comic fool p 266 Nevertheless Weizenbaum states three kinds of uses to which he feels on moral grounds computer ought not to be applied These are o projects that seek to substitute machines for parts of living animals to create hybrid bionic systems for example coupling an animal's' brain and visual system to computers Such projects are obscene and represent an attack on life itself One DUlst wonderwhat conceivable ' need of man could be fulfilled by such a 'device' at all p 269 o projects involving the substitution of a computer system for a human social function The use of computers or machines of any sort to take the place of a human participant in a function that involves interpersonal respect understanding and love is also obscene and ought not to be attempted Weizenbahum maintains because respect understanding and love are not technical problems pp 269-270 The obvious example of such projects is the use of computers as psychotherapists o projects that may clearly be seen to have unpredictable and irreversible side effects As an example of this kind of application Weizenbaum cites the automatic recognition of human speech I have no objection to serious scientists studying the psycho-physiology of human speech recognition he says the work he objects to is mere tinkering aimed ultimately at facilitating electronic snooping and Wiretapping ' He specifically mentions ARPA-funded speechresearch as the sort of work he means to condemn p 271 Weizenbaum ends his book with a chapter which honestly and eloquently states his own personal position Whether or not one agrees with anything he has said in the book I feel that this frank personal statement deserves the greatest respect He refers to his work as a teacher with which he began and which is obviously of great importance to him much to his credi t He says that much of his message is directed in particular to teachers of computer science Such a teacher no more nor less than any other faculty member is in effect constantly inviting his students to become what he himself is If he views himself as a mere trainer as a mere applier of 'methods' for achieving ends determined by others then he does his students two disservices First he invites them to become less than fully autonomous persons Second he robs them of the glimpse of the ideas that alone purchase for computer science a place in the university's curriculum at all p 279 Weizenbaum makes a final strong statement of his position The computer is a powerfUl new metaphor for helping us to understand many aspects of the world but it enslaves the mind that has no other metaphors and few other resources to calion The teacher must teach more than one metaphor and he must teach more by the example of his conduct than by what he writes on the blackboard He must teach the limitations of his tools as well as their power p 277 What should our reaction be to this book Admittedly it is often a polemical biased and emotional presentation On the'other hand many excellent and telling substantive points are raised whil h merit a serious effort at understanding on their own terms and a reasoned reply based on such an understanding The clear well-presented chapters on the nature of computers programs theories and models etc written by a man who has paid his dues through many years of competent and creative computer science research are in themselves rewarding features for any reader aside from the value judgments elsewhere in the book Dismissing Weizenbaum's positions out of hand as New Left propaganda is not in my opinion a justifiable approach to the book's more controversial points After all the deeper issue is not the particular computer applications he personally objects to but rather the fact that he challenges each of us to take up our own stand about how technology should or should not be used and what kind of a world we want for ourselves and our children Several critiques have already been directed at Weizenbaum's book by members of the AI community In the June 1976 issue of the SIGART NewsZettep a publication of the Special Interest Group on Artificial Intelligence in th Association for Computing Machinery pp 4-13 there is a highly interesting collection of May 77 CRYPTOLOG Page 24 UNCLASSIFIED - - -_ - _ _ _---_ - DOCI-o -- -_ _---_-- _-------'--- UNCLASSIFIED reviews and a rebuttfil by Weizenbaum all originally available on the ARPANET on-line communications network linking many major research centers The first review by Benjamin Kuipers of the AI Lab at MIT is entitled Reactions to Weizenbaum's Book It sets a very balanced and restrained tone and seems open-minded and even sympathetic to many of W izenbaum's viewpoints Kuipers clearly seems to understand and acknowledge Weizenbaum's frame of reference and agrees on the importance of avoiding a narrowly instrumental approach to human reason and scientific undertakings He claims however that AI researchers are already well aware of and engaged with these issues and he is puzzled by Weizenbaum's vehement attacks on the AI community He also accuses Weizenbaum of obscuring his valid points by harsh and sometimes shrill accusations and personal attacks p 4 A very long review by John McCarthy of the Stanford University AI Labor entitled An Unreasonable Book takes a much more impatient and hostile attitude pp 5-10 McCarthy characterizes the book as moralistic incoherent and self-contradictory on matters Weizenbaum has not thought through as carefully as he should have In several places he raises the interesting point that Weizenbaum's moralistic strictures may be twisted and exploited by activist bureaucrats or public interest organizations as an excuse to stifle scientific freedom or to redirect scientific endeavor I am frightened says McCarthy by the book's arguments that certain research should not be done if it is based on or might result in an 'obscene' picture of the world and man Worse yet the book's notion of 'obscenity' is vague enough to admit arbitrary interpretations by activist bureaucrats p 5 It seems to me in reading McCarthy's comments that science itself represents an ul timate value source for him like a religion He appears to accept it as such unquestioningly to the point where he is completely unable or unwilling to entertain the suggestion that any other value source might take precedence over science In support of his overriding respect for science he presents an uncompromisingly absolute and emotional statement by Andrew D White first president of Cornell University 'In all modern history interference with science in the supposed interest of religion has resulted in the direst evils both to religion and science and invariably and on the other hand all untrammelled scientific investigation has invariably resulted in the highest good both of religion and of science' p 9 For McCarthy evidently scientific freedom is an unquestioned value in itself which must be preserved as our first priority and at any cost McCarthy takes issue in his review with almost every point that Weizenbaum made in my opinion however he does not come to grips directly with Weizenbaum's essential arguments He dismisses themin o to D1Uchas many AI workers dismissed l Dreyfus' earlier work His approach is 'purely pragmatist anything is moral so long as it works Using computer programs as psychotherapists would be moral if it would cure people p 7 Weizenbaum's response to McCarthy forms the third and final portion of the presentation in the SIGART Newsletter pp 10-13 He reiterates the main arguments of the book and defends himself against some of McCarthy's more barbed attacks He provides a direct rejoinder for example to McCarthy's pragmatic justification for computer psychotherapy this reply can serve to bring into focus the basic gulf between the two viewpoints Prefrontal lobotomy 'cures' certain mental disorders But at what price to the patient and I would add to the surgeon as well I believe that machine-administered psychotherapy would induce an image of what it means to be human that would be prohibitively costly to human culture In closing I would like to single out three points that seemed most important to me in Weizenbaum's book These are at least in part highly relevant to many of our concerns as employees of a large government organization that uses computers o A solid theory base is crucially important The lesson of DENDRAL and MYCIN pointed out by Weizenbaum has been presented by others also the cooperation of one or more specialists who are highly skilled and gifted in the problem area and can transfer their knowledge into the program's knowledge base is a vital prerequisite for a successful application This means that the problem the data the habits of working of human problem-solvers the function of the automated process within a larger context etc must all be intensively studied and deeply understood Then a model must be developed which is sufficiently detailed and representative of all aspects of the problem situation to satisfy the needs of those who will use the program And ultimately this means more emphasis on learning about the problem the data and the man-machine interfaces and less emphasis on instruments and techniques as such more on the what and why and less on the how o Control ove ' the technology or profJ1'aIn is essential Someone or some small group of people must maintain a grasp on how the program or device performs Someone must be watching to see that the data and the problem situation do not drift away from the program's model Someone must be able to detect malfunctions and correct them not with a patch but by amending the model and if necessary also the theory beneath the model We must even be willing to junk the model altogether if it breaks down and build another All of this again forces us to direct constant attention to the problem the data and the human users who must work with the program May 77 CRYPTOLOG Page 25 UNCLASSIFIED DOCln --- - tf UNCLASSIFIED o Human goals have overriding importance We cannot take our values and goals for granted as if they were somehow part of the package containing our current scientific theories and the state of the art in some technology We must find a way to choose where to put our effort which programs to write which techniques to use which problems to solve -- a way that does not simply reflect the monetary or political pragmatism of the moment I am doubtful about the reality of pure research most research gets done because of passing fashions interests and demands simply because a given field is active and paying off at the moment or even worse just to get another paper into the literature I agree with Weizenbaum in wishing that basic human considerations could have a larger part in the motivation of those choices that will inevitably be made Weizenbaum does not tell us how to make better choices but he has had the courage to tell us how he himself has chosen what he will and will not do with computers References Buchanan B Sutherland G and Feigenbaum E A Heuristic DENDRAL A Program for Generating Explanatory Hypotheses in Organic Chemistry in Meltzer B ed Machine Intelligence New York American Elsevier 1969 Dreyfus Hubert L What Computers Can't Do New York Harper and Row 1972 Martin W and Fateman R J The MACSYMA System in Proceedings of the Second Symp- sium on Symbolic and Algebraio Manipulation New York Association for Computing Machinery 1971 Mumford Lewis Technios and Civilization New York Harcourt Brace Jovanovich 1963 Ornstein Robert E The PsyohoZogy of Consoiousness San Francisco W H Freeman and Co 1972 Simon H A and Newell A Heuristic Problem Solving The Next Advance in Operations Research in Operations Researoh Vol 6 January-February 1958 pp 8 ff Weizenbaum Joseph Symmetric List Processor Corrmunications of the ACM Vol 6 No 9 September 1963 pp 524-525 ------ ELIZA -- A Computer Program for the Study of Natural Language Communication Between Man and Machine Ibid Vol 9 No 1 January 1967 pp 36-45 ------- Contextual Understanding by Computers Ibid Vol 10 No 8 August 1967 pp 474-480 Winograd T Procedures as a Representation for Data t n a Computer Program for UndErstanding Natural Language PhD Dissertation submitted to the Department of Mathematics MIT August 24 1970 i $ From Weizenbaum's Prefaoe I A number of practicing psychiatrists seriously believed the Briefly summarized I composed a computer program with which DOCTOR computer program to have the potential of growing into one could converse in English by means of a typewriter a nearly completely automatic form of psychotherapy I chose the name ELIZA for the language analysis program be2 I was startled to see how quickly and how very deeply cause like the Eliza of Pygmalion fame it could be taught to speak increasingly well people conversing with DOCTOR became emotionally involved with the computero My secretary who had watched me work On For my first experiment I gave ELIZA a script designed to the program for many months and therefore surely knew it to be pennit it to play -- I should really say parody -- the role of a merely a computer program started conversing with it After Rogerian psychotherapist engaged in an initial interview with a only a few interchanges with it she asked me to leave the room patient DOCTOR as ELIZA playing psychiatrist came to be known soon became famous around the Massachusetts Institute 3 Another widespread and to me surprising reaction to the of Technology and at other institutions in the United States The program became nationally known and even in certain ELIZA program was the spread of a belief that it demonstrated a circles a national plaything general solution to the problem of computer understanding of natural language In my paper I had tried to say that no general The shocks I experienced as DOCTOR became widely known solution to the problem was possible i e that language is and played were due principally to three distinct events understood only in contextual frameworks May 77 CRYPTOLOG Page 26 UNCLASSIFIED DocrD 4009794 UNCLASSIFIED TEl r 263 P L 86-36 Reprointed from C-LINERS Feb2'WlPy-March lB 5 '- - T Boy Just think abo t 50-words describing shades of blue his technique will eschew the obfuscation At this point I am about to be accused of prevailing in the system When I first read among other things being against mother lov this line it was Monday morning and I hadn't apple pie and hot dogs but really I'mnot I' had my first coffee In my excitement fumbling love Mom for the phone under emergency conditions I spilled my coffee It ran under my Webster on Who except the one trained in chromatography the desk Still grappling for the phone I can distinguish between ochre and saffron both pulled the dictionary deeper into the coffee In shades o ye lo Possibly he's the sam person spite of all this my only thought was to make who can dlstlngulsh between use and utilize my call and get rid of the obfuscation in the or possible and feasible ' system Even the job titles of our duties give The clue was right on my desk The dictionary credence to this attempt to cushion the impact was now coffee-soaked It was as though Old Noah of our communication What does a coordinator was deliberately maneuvering his book into a do What is a focal point In the latter position which would prevent my using the phone title I envision many inputs being channeled But I still didn't catch on through a small opening sometimes called a lens ometimes a bottleneck and emerging inverted ' Wouldn I t you know it The line is busy completely reversed at some secondary destinaPerhaps a fast rereading of the memo would help tion Is a focal point a passive position in me to better understand the problem Yes this which the only requirement is to sharpen up the guy was in real trouble Not only was he image of a previous input as does a lens or plagued by obfuscation but he also had a can it take independent action and contrib te to plethora of common recordings I wonder why the output My apologies to all who must bear he didn't call me instead of writing a memo up under this label Perhaps he also had a Webster on his desk It is all too easy to poke fun at ourselves Now this has gone far enough and I apologize to my readers but I cannot exonerate Webster for but it takes a little more thought to find solu-' tions So then why do we choose words for a his part in this fiasco of communications In his desire to record our language Webster unwit- particular job anyway There are many reasons you mayor may not agree One textbook says tingly gave us an overabundance of choices of that we write to direct inform or persuade I words One might say that we have the Midas have observed that we also write to impress touch in our lexicon We like the king may confuse stall or even retaliate Then there starve to death unless we discipline ourselves is the neutrality syndrome Some writing is in the art of choosing done of all things to communicate Perhaps Further compounding the problems are the many you could make a few additions to my list thesauruses lying around unguarded like Pandora's If upon weighing a particular reason for boxes just waiting for the unwi tting to lift the writing the decision is to communicate I cover Now I did it 300 words to replace use believe that there is but one rule in picking but only 50 to replace possibility but each the best word pick the shortest and most common you must remember is just a shade different from the others Leonardo would certainly have word Risk being ualled a square Make the stopped with finger paints if someone had told decision now Decide whether you would rather eschew the obfuscation or avoid the confusion him that there are over a hundred shades of brown Then what about Gainsborough's Blue s 55 S May 77 CRYPTOLOG Page 27 ONCLASSIFIED ' ' '''''' DOCID 4009794 UNCLASSIFIED CRYPTOLOG ApM Z 1977 L R Chauvenet by P12 I I The - Th - ru-st' s--- a-n ' d-- 'Bo Jgg1es 0 f Outrageous Escalation NSA TeahniaaZ JOUI'naZ P L ANNDUNCEMENT SOZution to NSA-crostic No 7 Vol XIII Spring 1968 No 2 Not everything has to be per se in order to be real reasonable deals were rarely mentioned but the old quid pro quo was much in evidence but the true winner in the forergn competition was vis-a-vis a beguiling French ' sound UNCLASSIFIED _II_ - o_ '-''-'_ _ _II'''''_ ''' _O_'''-' _ _II'''I To the Editor CRYPTOLOG I must confess that I thoroughly enjoyed -even found hilarious -- the article in CRYPTOLOG entitled A Medal for Horatius JanuaryFebruary 1977 I see the colonels' club was active in those days too Having had a similar experience in the not too distant past -- trying to justify the awarding of a Legion of Merit decoration to one of my military subordinates over the objections of various but nameless peoples and boards -- I found it gratifying to realize that inanity can also be humorous Thanks for the laugh '-- With this issue lends her tour as Cryptanalysis Editor of CRYPTOLOG Betty has been a member of the Board of Editors since even before the first issue of CRYPTOLOG appeared in August 1974 In addition to contributing articles of her own What Should You Expect or The Analysis of Cryptanalysts August 1974 Secrets of the Altars -- The Moustier Cryptogram September 1974 An October Overlap October 1975 ind Twenty Years of Transposition August 1975 Betty has been of invaluable aid in encouraging cryptanalysts young and old to write articles forCRYPTOLOG In her current assignment however Betty has fewer opportunities to sniff out good articles and therefore has asked to give this important job to someone else who has more such opportunities I IG42 has graciously agreed to become Betty's successor as Cryptanalysis Editor and promises to beat the bushes for good articles 11 that field So Betty thanks for doing a wonderful job all these years and Alice now it's your turn -- so go get 'em William Lutwiniak Publisher __ _ _ _ __ _ - - ------ - - - - JIB311 P S Thanks to Regular Army channels the award finally came tbrough UNCLASSIFIED _ __ I _ _ '- _II_ O_II_ _II_II __ _ _ _'_ - - DID YOU GET YOUR INDEX The cumulative index 1974 through 1976 of CRYPTOLOG articles and authors' names was distributed in February according to a special distribution list If you requested a copy but did not receive one or if you did not request a copy originally but would like to have one anyway please write to CRYPTOLOG PI or telephone the editor on 5236s UNCLASSIFIED _ _ _ __ - _ _ __ _11 _11_ _ __ 0_ May 77 CRYPTOLOG Page 28 pl-Apr 77-83-25237 UNCLASSIFIED 86-36 I This document is from the holdings of The National Security Archive Suite 701 Gelman Library The George Washington University 2130 H Street NW Washington D C 20037 Phone 202 994-7000 Fax 202 994-7005 nsarchiv@gwu edu