nm LI E> R.AFLY OF THE •"*"- UNIVERSITY Of ILLINOIS 510. 84 *©.324-33d cop., 2 », The person charging this material is re- sponsible for its return on or before the Latest Date stamped below. Theft, mutilation, and underlining of books are reasons for disciplinary action and may result in dismissal from the University. . University of Illinois Library ;•:•/ ^ 1 M 4 1571 OEC 1 9 RE* FEB - 1 \m JAN 2 5 MAR 28 197 APR 2 i SEP 3 1977 fEfi LOHtC'Q MR 1 4 i9p 8 "M i o rati Ml ttui MAY 2 9 IpT OCT 2 1 WO NOV 1 4 1970 NOV 1 OCC - 7 WO 2004 MAR 8 1976 MAR 8 RECD ■ L161— O-1096 194 Digitized by the Internet Archive in 2013 http://archive.org/details/roleofuniversity330past PI" so Report No. 330 THE ROLE OF THE UNIVERSITY IN THE COMPUTER AGE by JOHN R. PASTA May 20, 1969 Report No. 330 The Role of the University in the Computer Age by John R. Pasta May 20, 1969 Department of Computer Science University of Illinois Urbana, Illinois 6l801 THE ROLE OF THE UNIVERSITY IN THE COMPUTER AGE 1. Introduction In any description of the Impact of computers on universities it is important to separate the role of computers in the research and educational processes of the institution from its status as an object for study. This distinction is not always made carefully and, indeed, it is not an easy one to make as can be realized from observing the many disagreements over the definitions of the discipline Computer Science. It is agreed generally, how- ever, that there is a rapidly growing body of knowledge which lies outside or, at least, on the fringes of the usual university disciplines which displays a coherence sufficient to call it a discipline in its own right. The most common name for this area is Computer Science. In the areas outside of computer science proper, computers have played an important role in both the research and educational functions of the uni- versity and this position is becoming more important each year. These relation- ships will be examined in order to gain an appreciation for the overall position of the digital computer in modern education, but it is the consideration of the growth of computer science in the university which will be treated first. Some consideration will be given to why computers have become an object of intense study. It is fashionable, these days, to re-examine educational goals in terms of relevance to society and to life. There can be no denying that these new machines will have a large secondary impact on the living and working habits of our society but with very little extension in thought one can imagine awe-inspiring and frightening consequences of present-day developments -2- in the science. We are accustomed to developing our science and reserving until later consideration of the consequences, a habit which, one day, may be our undoing. 2. Early Contributions The origin and early development of computers was carried out main- ly in universities. The pioneering work at Harvard University, the University of Pennsylvania, the Institute for Advanced Study and in England was picked up quickly at a dozen places . The potential of the digital computer was recognized and this early work demonstrated that electronic components had reached a degree of reliability permitting the fabrication of systems of the size contemplated. In addition to important research contributions such as magnetic core memories, index registers, assembly programs, and compilers, the univer- sities produced the engineers and mathematicians who were responsible for the spectacular advance of computers in the first decade. In those formative years the spirit was more cooperative than competitive. In the best university tra- dition, the communication of ideas, progress and results was given an important position. In reviewing a report from the group at the University of Illinois 2 M. Lehman of the IBM Corporation wrote: "The history and valiant pioneering efforts of the Digital Computer Laboratory of the University of Illinois are perhaps not as widely known and acknowledged as its many fundamental contri- butions to the development of computer art and science warrant. In fact, with the regretful but inevitable transfer of computer hardware research and development from the university to industrial surroundings, this laboratory is almost the sole survivor of the -3- many groups who, for a decade and a half, conceived and/ or laid the foundations of the entire field of information processing. "Among its many contributions, Illinois can pride itself on its record of documentation. There is probably no other machine whose development and final design is so completely documented and described as the ILLIAC II. This machine, at its conception in the mid 1950' s, represented, together with some other independent design projects of the same period, the spearhead and breakthrough into a new generation of machines . The early history of computers and com- puting is becoming rapidly obscured, and it would seem urgent that some historian attempt to reclaim and record the facts while the men who made them are still around. Such a researcher would certainly include among his basic documents the series starting with the now classical Report 80 of the University of Illinois' Digital Computer Laboratory, so as to trace the early history of concepts, e.g., modu- lar interleave memories, instruction look-ahead, redundant number representations, and computer arithmetic in general." 3 . The Computer Explosion and its Impact on Computer Science Even while the earliest research was underway there were beginnings of a tooling-up by industry for the great proliferation of computers to follow. It was not long before there was little justification for a university to build computers to satisfy a service need at the institution. Unfortunately, it was this service need that had been advanced as a justification for most projects in computer hardware research because this procedure insured the broadest -based institutional support. Not everyone viewed the role of these early computer scientists in -k- that light. ■ In an early confrontation between a university group and a short- sighted industrial management A.E.C. Commissioner John von Neumann asked "where shall we get the engineers and innovators of the future if these acti- vities are disbanded?" This view did not prevail, however, and university interest in the area moved away from the pragmatic aspects toward the more theoretical consid- erations which underlie the discipline. It was this shift in emphasis which may be responsible for the evolving in most curricula today of "Computer Science" rather than "Computer Engineering." These two aspects are not incompatible in the discipline. Neither are they incompatible in an individual and the prime example is to be found by examining the accomplishments of von Neumann. There is, in fact, a tendency these days to try to reintroduce into the curriculum at least some of the systems architecture ideas which implement the body of abstract knowledge being assembled. Of course, the earlier concerted drives to con- struct large general purpose computers at the university have disappeared, but the need for some of the activities associated with machine design and fabri- cation remains. It has long been recognized that consideration of computer systems must embrace hardware and software simultaneously for a properly integrated picture. The deeper studies of these systems can not do otherwise if one is to have a true perspective. k. Education in Computer Science Curricula in Computer Science have been adopted at most major uni- versities not only because of faculty inclination but also to meet the needs of students and society. Although the initial programs reflected the special interests of the charter faculty, the departments enlarged and diversified to -5- cover the discipline in a nearly uniform way. The general pattern of such a curriculum became sufficiently clear that various organizations and committees could undertake the job of defining it in general terms. These attempts were meant to make models around which a program suitable to the local university environment could be formed. There are many ways of categorizing the different components of a curriculum but the classification of the Curriculum Committee on Computer Science of the Association for Computer Machinery enjoys the best professional 3 standing. The three major divisions with their principal topics are: I. Information Structures and Processes 1. Data Structures 2 . Programming Language s 3. Models of Computation II. Information Processing Systems 1. Computer Design and Organization 2 . Translators and Interpreters 3. Computer and Operating Systems k. Special Purpose Systems III. Methodologies 1. Numerical Mathematics 2. Data Processing and File Management 3- Symbol Manipulation k. Text Processing 5. Computer Graphics 6. Simulation 7- Information Retrieval -6- 8. Artificial Intelligence 9« Process Control 10. Instructional Systems In order to give the reader a notion of the content of these areas they will "be treated "briefly. 5 • Information Structures and Processes One of the attractive features of mathematics is its suitability for abstracting essential features from complex situations and dealing "with them in a rigorous and elegant way. There appears to be a low threshold for the degree of complexity found in nature which the mind can handle. The abstracting procedure and the establishment of relationships among these attributes is really what is meant by "handling" a complex situation so we really ought to define more carefully what we mean by degree of complexity in order to understand its reduction. Nevertheless, the reader will have some feeling for these matters; the relative complexity of, say, chess and checkers is something he will accept without requiring a quantitative measure. The digital computer is an instrument designed to handle complex situations and as the technology advances the potential situations become more and more complex. The handling of the payroll for a large corporation is clearly a complex affair, but while the data base for a payroll may be complex, the procedure is straightforward. What we might consider is a procedure so complex that it must be the product of many minds and is incapable of detailed comprehension by any one person. We have instances of such computer programs and the analogy with attaining a detailed knowledge of the working of the brain should be evident. -7- In order to specify these computational systems, however, the computer scientist must understand his system and its data at a level of low complexity so that he can find basic theorems or universal truths to guide him in the proper design of the overall systems and data bases . In using the term "system" we encompass, as is customary, both hardware and software aspects. The courses in this part of the curriculum prepare the student for an understanding of the representation and manipulation of data and programming languages in various organized ways . The study in depth of these structures has led to theoretical models and to the more abstract parts of computer science. Models of computation lead one to studies of switching circuits, sequential machines, automata theory, formal languages and grammars, mathematical theories of languages and of computation. As an example of the kind of thing we are talking about, consider the Turing machine, a model invented in the 1930* s by the mathematician A. M. Turing. This abstract model is very simple. In one form it is a device with a finite number of internal states and a tape of arbitrary length marked into squares. At any moment it can read a symbol on the tape. Based on that symbol and the internal state, the machine can initiate actions to change the symbol and move the tape one square left or right. One would expect such a machine to be limited in the kinds of things it could do and yet Turing showed that any effective computation performed on any computer can be performed on a Turing machine. The universality of this machine allows us to establish truths about it which will apply to all other machines and consideration of this and other equivalent models has increased our understanding of computers, programs, languages, and computations, all of which can be fit into this simple model. -8- Indeed one may ask another question. Does such a machine have enough "intelligence" to build a copy of itself; to reproduce itself? Offhand, one might think that an object would not build a replica with the full complexity of the constructor. We do, of course, have examples in nature if we accept a mechanistic view, and we can, in fact, show that it is possible in the case of these machines. It is the possibility of considering a question such as this that make these models so interesting to researchers in computer science and particularly to the student seeking a deeper under- standing of computers and more general automata. Thus we see that this part of the curriculum and of research gives rise to a kind of intellectually satisfying exercise for serious students of computer science. 6. Information Processing Systems In computer science a system is usually considered to be the whole complex of hardware and software with all of the connections, interfaces, and communication channels considered as a single entity. In the education process, however, the interconnection and interdependence of hardware and software is largely lost when the student moves from a hardware -oriented session to a software class. Of course, this phenomenon is not peculiar to this discipline and in many fields of study the making of the connection or integration into a full picture is left to the student. The tragedy is that it is often impossible for the student to fill in the hardware part of the total system picture even when the term "hardware" is used in the loosest of connotations. This part of computer science is frequently absent in a curriculum and may have arisen from the historical pattern mentioned earlier. -9- The best possible situation would be the one in which a faculty with competence in all phases of system study worked closely together in a single department. Computer science covers such a broad area, however, that only with the largest of departments may one hope to cover this field with reasonable depth. Furthermore, the broadness of the field brings together such diverse interests that a certain amount of separation must occur. In spite of the fact that we call the discipline a science, it should be realized that the heart of the subject is the engineering of hardware and software systems. This realization has led in some cases to the incorporation of the discipline into engineering departments. Anyone watching the spectacular growth of computer science can see easily that such a department with multiple goals must grow quickly beyond reasonable size and fission off a computer science department either formally or practically. The problem is rather the other way; as a computer science discipline grows, it may break up into subdisciplines and the department becomes a school or college. Whether the connection is made, however, it is now conceded generally that a well-rounded computer scientists will have a good knowledge of computer architecture, both hardware and software, at a fairly detailed level. Usually the level of detail for software is more atomic than for hardware. A recent aspect of study in this area has been the attention to the diagnostic study of complete systems in an attempt to design better systems or better utilize present systems. This is often carried out as a kind of behavioral approach in which the detailed structure becomes less important than the action of the system in different situations or the means by which it can be made to function better in these situations. -10- This second "broad area of computer science,, Information Processing Systems, is the hardware and software systems engineering part of the dis- cipline. The structure of this part of a degree curriculum is the most con- troversial with respect to both content and depth and these questions remain largely unresolved. 7- Methodologies The field has grown so rapidly that the different applications areas have not drawn far apart with their own methods, jargon, and intellectual organization insofar as computation is concerned. It is possible to extract the methodology common to the many applications of computers and study their structure . The oldest and the one with the largest "body of knowledge to draw from is numerical mathematics. The easiest extrapolation to make when seeking jobs for the early computers was from the desk calculator. This volume contains examples of the increases in the size and complexity of computations which the computer has made possible. Along with this increased capacity there has been a need for a finer measure of the meaning of results, for algorithms with better properties when applied to larger systems, as well as a basis for un- derstanding the mathematical stability of these procedures. In spite of the long history, numerical mathematics has barely kept pace with the new require- ments generated by the computer and the subject has enjoyed a renewed dynamism. Even such relatively stagnant areas as finding the roots of a function or approximating a function with other simple functions have provided arenas for exciting progress. This renewed interest in numerical mathematics is being overshadowed, however, by the growing research effort in non-numerical fields and the complexion of computer science is changing as more researchers, especially -11- recent graduates, move in that direction. It is hard to tell, at this time, whether this change is due to realization of its importance, the possi- bility of comprehending a field which has not yet accumulated great depth, the great diversity which thins competition, or pure huckstery. The manipulation of non-numeric information is' less completely understood and less formalized than is the handling of numeric information under the ordinary rules of arithmetic, even though most of the world's in- formation is of the non-numeric sort. The abstract study of data structures mentioned in the first broad educational category serves as a basis for techniques in processing library, medical, management, literary, military, scientific, ecologic, and other types of information files. These techniques would include organization, classification, and the vast problems of informa- tion retrieval on a large scale. A related topic is the formal manipulation of symbols according to fixed rules such as the processes of algebra or formal differentiation and integration. At the other end of the spectrum there has been an increasing interest in text processing and the humanities scholars have been carrying out programs in textual criticism, concordance generation, as well as metrical, 6 authorship, and. linguistic analysis at a high level of sophistication. At many places the humanists are engaged in exploratory research on textual material in advance of others one might have expected to provide the leader- ship, such as library scientists . 7 Operations Research found its first important application during the second world war and this area has expanded into a whole study of simula- tion in general. Business games, agricultural models, economic systems, transportation networks, university computer centers, and time-sharing networks all find their development better understood and enhanced by -12- simulation by mathematical models on the computer. .Special languages have been developed to deal with these situations and much computer time is devoted to this activity. Indeed, in a competitive situation a lack of such an activity or an inferior model could mean a failure of an organization whether it be a corporation, an agricultural complex, an economic system, an army, or a nation. There is always the danger of overselling an excessive dependence on models, but neither should inactivity lead to a waste of computer resources in po- tentially valuable studies. A characteristic of these simulation studies is their insatiable appetite for more and more computer capacity. Thus, consid- eration of the simulation and modelling areas is closely linked to the study of information processing systems. Automated factories have been considered for a long time and in many areas, such as the chemical industry, it has reached a very advanced state. The presence of the computer has changed the picture and opened the way for new approaches. From the simple feedback control of earlier systems there has been a progression of more advanced systems which are reacting in real time with complicated calculations of functions of the systems parameters . 8 One of the more promising directions is the automation of machine tools to perform operations not possible under human control. Feed rates can be computed in real time and surfaces can be generated with none of the classical restrictions to translatory and circular generation found in simpler machines under human guidance. The steps from specification to the final object is in the process of being completely automated, and the social implications are far-reaching. The Soviet Union has established an office of automation at the Minister level with the expectation that a breakthrough in this area will yield economic advantages over competitors who must develop highly skilled workers to approach the capabilities of future automated production. -13- q In the computer field itself, design automation has reached the point that layout and interconnection paths are generated "by the computer. A driving program is then written to control the plotter producing the art- work for masks which eventually will be used for printed circuit wiring and integrated circuit manufacture. Automation already has made incursions into the field of research particularly in chemistry, crystallography, nuclear physics, "biology, medicine, and other fields. The high intensity meson facility under construction at the Los Alamos Scientific Laboratory is an accelerator with a level of auto- matic control far above any now in operation. 10 The new field of computer graphics has produced a front end for the design procedure which allows a design engineer to manipulate visual diagrams, charts, and parameters, and in interaction with the computer, to develop the specifications for parts and assemblies. The power of this interactive mode, which is generally carried out with a cathode ray tube display, is not restricted to this type of operation, however. The adjustment of solution parameters in solving mathematical problems, the specification of computer programs by flow diagrams, the guidance of programs for recogniz- ing and measuring patterns of various kinds, the training of airline pilots, the behavior of models of physical systems, and the control of air traffic are a few possible applications of computer graphic techniques. There is much basic work in signal compression, low-cost displays, learning theory, image enhancement, and development of languages to he carried out in this wide-open research field. Computer graphics is also a facet of the problem of computer- based educational systems . ' Although this area lies mostly in the domain of education there are some points of contact with computer science. The -11+- tremendous impact and importance of this area can hardly he overestimated. The possibility of quickly raising disadvantaged individuals, groups, or even whole nations in an accelerated time scale to a new educational level, for example, opens a whole new dimension in this new world where technological sophistication has become as important as natural resources. The last defined area in the curriculum is that of artificial intel- 12 ligence, ' by which we mean a behavior of a machine which could be described as intelligent when observed in a biological system. This is a "fun" activity in computer science, and yet it is a most important one. It is not so much that computers should be designed to act in human-like ways any more than vehicles should have legs rather than wheels . It is rather that we can thereby better understand the organization and functioning of the mind. This is not meant to exclude the possibility that the things learned in this anthropo- morphic approach will not be applicable to the design of machines; the two are intertwined problems. The brain is made up of relatively simple components just as the computer, and the Gestalt of the mind may be thought of as the analogy of the computer program. We may learn some tricks from what appears to be a very professional job of programming in the biological case. Some beginnings have been made in this area, but much remains to be done beyond the present work, which is mostly organizational and descriptive. One of the odd characteristics of these early studies is the "magic act" nature of the results. A computer program, like a professional magician, can amaze and mystify with its apparent great powers, but when the details of the trick are 13 disclosed, we say, "Oh, of course!" The game of chess played by the computer is not of the highest level, but it is, nonetheless, quite impressive. A computer cannot perform by examining every move because of the astronomic size of such a task. Instead, it is programmed to assess possible moves -15- according to gains or piece -exchanges, resulting mobility, control of criti- cal squares, or other rules designed to achieve specific goals and subgoals. In other words, some of the chess experience of humans is taught to the machine The point is that with relatively simple criteria some surprising behavior can be generated; surprising, that is, if we are ignorant of the program or the program is too complex to comprehend. Ik Another area of intelligent behavior is pattern recognition and this area has received much attention. It is one in which it is fairly easy to determine the degree of success of a design, and at the same time will have important applications. It is also one, however, in which hardware system overtones are present and progress through software research alone is severely restricted. Overlying all of these efforts is the realization that true intel- ligent behavior must involve learning or adaptation and some of the deeper studies are in that field and it is here, the author believes, that lack of hardware support has been the most crippling. The lack of support is not just in funding but in the unavailability of computer scientists with hardware competence. The reader may be struck with the remarkable richness of computer science and the interdependence of the component parts. Coupled with the diversity and the newness of the field is a thinness in certain areas, but this just means a wealth of opportunity for significant research. 8. The Computer as a Tool The remainder of this volume is concerned with the application of computers in science and technology so there is no need to elaborate on that aspect here, but it is appropriate to mention how the mechanics of -16- using computers has evolved on the campus. When the first computers arrived on campus they were rudimentary and required considerable experience to use effectively. There were few users and even personal interaction with the computer was often possible. As the use became more widespread and higher level languages appeared, the computers grew in size and a large resident "system" came into being to handle the traffic. The consequent loss of personal interaction with the system was felt keenly, particularly in the preliminary checkout phases of computer pro- grams. Those who did not need the massive power of a large computer considered the possibility of a smaller, more personal system. Time -sharing systems, which in effect made the large computer look like many smaller ones, each serving a personal console, sought to satisfy these needs. A third approach was to establish a minimum local system to process immediate needs with a link to a larger backup computer. Although all of these systems have their drawbacks, it is a fact that many campuses incorporate all of these methods with the result that systems are becoming more dispersed. This has placed a premium on the study of computer system organization in an attempt to reintegrate the components into a more powerful whole which can operate more or less as a unit when required to do so, and yet preserve the integrity of the parts. How will computing power be dispersed? It might be through the utility concept of a large central facility with simple access, or perhaps a network of equal partners with symbiotic capabilities. More likely it will be a network with a hierarchy of computers of different capacities. Whichever it is to be, it appears that the university campus is a fertile ground for testing these ideas. -17- 9« Concluding Thoughts There has "been no attempt at describing specific research projects at universities. These research activities can be deduced from the outline of the curriculum; the teaching and research at a university merge imperceptibly from one to the other in the field of computer science. In any event, these projects change very rapidly in this new discipline and one can only conjecture on their direction. Consider one of the most dynamic and intellectually challening fields, that of artificial intelligence. It is likely that the overlap with the bio- logical sciences will increase and the consequences of such an interaction could be startling. The unravelling of the structure of DNA, the primary genetical sub- stance, was a monumental task involving the analysis of a really complex structure made up of constituents obeying simpler laws. More and more biolo- gists who have worked in this area are turning to an analogous problem, the structure of the human brain, also made up of much simpler components, and they are being joined by members of other disciplines. In many respects this latter problem is more germane to study and alteration of the human condition than is biological evolution. It is now generally felt that environmental conditions and evolution of a psychosocial type are the more significant factors molding human life today and that the biological evolution of man has reached a virtual standstill over the last 100,000 years. The biological constitution of the man of the twentieth century and of prehistory are not significantly different. Man carries in his brain rather than his genes the accumulation of recent evolutionary changes and these are passed on as surely as imprints on the genetic material. If this thesis be accepted, then there is a strong argument for a -IB- study of how man's "brain has been programmed "by external forces and this problem looms larger than the genetical one. With technological advances accelerating at the present rate the biological functioning of the human body is becoming an anomaly and the furnishing of mechanical substitutes for the more important constituents is rapidly becoming an important business. 16 There have been proposals for life support systems for an isolated head and technical problems may not be unsurmountable . This may seem a macabre proposal, but it is a logical extension of current survival systems. Another feature of a social evolution theory is the control of the environment by man and the closing of the loop by the effect of the environ- ment on the controlling agency. A better understanding of this cycle may veil be essential to man's survival of technology and prevention of a mental ex- tinction. These considerations lead to an interest in brain mechanisms and one of the best approaches currently available is through use and study of the computer. This is a single example of a need for the study of computers for their own sake as well as with a view toward applications. It is difficult to imagine much of this kind of research outside of the university. Fortunately it is being carried out at universities, although probably more as a result of the intrinsic intellectual interest rather than owing to a strong social consciousness. Nevertheless, the impor- tance to society ought to be recognized at some level in the decision process. There is a need for informed opinions in all phases of this new science if its impact on our future proves to be even a fraction of informed expectation. -19- References 1. S. Rosen, "Electronic Computers: A Historical Survey." Computing Surveys 1 (1969), pp. 7-36. Note the editorial comment on p. 2 of this issue, however. 2. M. Lehman, Review 9119. Computing Reviews 7, I966, pp. 83-84. 3. ACM Curriculum Committee on Computer Science, "Curriculum 68," Communications of the ACM 11 (1968), pp. 151-197- k. M. L. Minsky, Computation : Finite and Infinite Machines ♦ Prentice- Hall, Englewood Cliffs, N. J., 1967, pp. 103-1^5. 5- G. Estrin, et al., "Snuper Computer - A Computer in Instrumentation Automation," Proc . Spring Joint Computer Conference , 19&7* Thompson, Washington, D. C, 1967. 6. E. A. Bowles (ed.), Computers in Humanistic Research , Prentice -Hall, Englewood Cliffs, N. J., 1967. 7- S. Beers, Decision and Control , Wiley, New York, N. Y., I966. 8. 0. S. Puckle and J. R. Arrowsmith, An Introduction to Numerical Control of Machine Tools , Chapman and Hall, London, 196k. 9- Proceedings of the SHARE Design Automation Workshop , June 23-25, I965, Atlantic City, N. J. 10. D. Secrest and J. Nievergelt (eds.), Emerging Concepts in Computer Graphics , Benjamin, New York, N. Y., I968. 11. See report of seminar at University of Texas, Austin, Tex., Computer - Assisted Instruction , Testing and Guidance . Harper and Row, New York, N. Y. (to be published October, 1969). 12. E. A. Feigenbaum and J. Feldman (eds), Computers and Thought , McGraw-Hill, New York, N. Y., I963. -20- 13. E. A. Feigehbaum and J. Feldman, loc. cit ., pp. 37-105. 1^. G. G. Cheng, et al. (eds.), Pictorial Pattern Recognition , Thompson, Washington, D. C, 1968. 15. P. B. Medawar, The Future of Man, Basic, New York, N. Y., i960; G. Wols- tenholme (ed.), Man and His Future , Little, Brown, Boston, Mass., 1963. 16. N. Amason, "Brain Without a Body?", Soviet Life (September, 1968), pp. k6-l&.