The Machine If we accept the premise that all computing gravitates to the largest possible machine, small machine proliferation becomes untenable. While this does not preclude the use of small special-purpose devices for special tasks, it does seem to preclude the need for each user group to have its own machine and appropriate staff. An unfortunately common first step into the computing field is the acquisition of one or more small machines with complete open-shop operation. While this type operation is rewarding to the user, it is somewhat difficult to justify on a cost basis, and the user soon becomes disillusioned by the limited size and speed of the machine. At this point most of the users are willing to forego some of the freedom of small machine operation in order to acquire size and speed. The next step is the acquisition of a large high-speed machine to be operated in the batch processing mode. In most instances the small machines remain. Thus begins the migration from machine to larger machine, in an effort to get the computing capability needed. Each machine change leads to an anguished period of problem restatement, reprograming, and reorientation. Each machine change brings glee to some and pain to others. One cannot start with the small system and go to the large system without major upheavals or major discontinuities. The third generation of modular computers with their building-block design and complete upward compatability may make future increases in computing capacity less painful. If we can assume that the present concepts of a modular computer are to be in vogue for a reasonable length of time, then a progressive plan with incremental steps can be outlined in such a way as to reduce the alternate feast and famine of computational capability brought about by the discrete computers of the past generations. The advent of the time sharing systems with the provisions for terminals, shared memories, shared peripherals, shared processors, and teleprocessing, should permit the university to lease or purchase just the amount of computing needed and be able to react quickly to changing requirements. When the terminal is mentioned, one normally visualizes a typewriterlike device with someone operating the keyboard at a poor typing rate. The term terminal should be taken to mean any input-output device available. One can visualize not only typewriterlike terminals in using areas, but also high-speed readers and printers, graphical display devices, and small peripheral computers which store data and process it to some extent but call on the central computer for difficult processing and computation. Such terminals, and problem-oriented languages and compilers open a whole vista of possibilities for university computation utilizing a central processor and time sharing. APPENDIX F WHAT COMPUTER FACILITIES ARE APPROPRIATE The Larger College or University A college or university large enough to utilize the computational capacity of a typical computer system would probably decide to operate such a system within its own walls. In the past, this has been supported by a combination of manufacturers' discounts, a Federal grant (usually from the NSF), and funds from the university itself. In some cases complex arrangements with manufacturers have been worked out so that the institution itself bears a relatively small portion of the total cost. In most institutions, the presence of Federal grant money for related projects is used, and in some cases actually sought, to offset part of the operating expense. The extent to which this additional support is available will vary widely, but it cannot be expected in the future to be used to offset costs of educating students. Furthermore, it would be difficult for many predominantly teaching institutions to attract the amount of research funds needed to make any sort of dent. Thus, it would appear that the methods by which colleges and universities have equipped themselves with computers cannot be counted on to supply future educational needs on a broad scale. Smaller Colleges There are two ways in which smaller colleges can begin to provide computing power for educational needs. The first is through the acquisition of a "small" computer, such as has been done quite often in the past with the help of matching grants from the NSF Undergraduate Instructional Equipment Branch. This has, in the past, been an effective means for introducing computing to a large number of small colleges, those that were able to present a convincing argument to the National Science Foundation. The small computers provide a good mechanism for the training of a corresponding small number of students, many of whom have been in the past headed for further training in computer science. However, the inability of a small computer to present a truly sophisticated software system for the user will prevent this method from becoming an important vehicle for mass training and indoctrination in elementary programing and principles of computing. Further, the financial burden of this course is high. Manufacturers' dis counts have been lowered, and the housing of a computer may pose significant financial hardships on many colleges. Smaller Colleges Perhaps the most sensible way for small colleges, and many larger institutions, to provide educational computing service to their students is to obtain one or more teletypewriters or similar consoles connected to a very large and very sophisticated computer system through telephone or telegraph lines. This approach has several advantages. First, the amount of computing power to be supplied to a given school can be easily tailored to the amount of funds that are available. Second, the institution does not have to assume the task of administering, and sometimes developing, a large-scale computer system. Third, every student has the advantage of being able to call on the most sophisticated software systems, something that most institutions could simply not supply on their own. Actual experience has shown that a single teletypewriter can expose computing to hundreds of students during the course of an academic year. Naturally, a deeper involvement in computers with more frequent exercises will require additional teletypewriters. One small college has actually used this method to get started in computing. Harpur College in Binghamton, N.Y., has been connected to the Dartmouth Time-Sharing system for almost a year. They plan to obtain their own large computer, but would also be quite happy to hook into a "New York Educational Data System Network." The ability of a time-shared computer system to provide computing power flexibly, quickly, and in a wide range of quanta, permits a wide variety of administrative structures for providing this service. Large schools, of course, can have their own private computer system. They would be able to experiment with various aspects of computer science and should do so. Components of a state university system could expect to obtain whatever specific computing they needed by tapping into a statewide network. Secondary Schools The argument for obtaining service from a time-shared computer system is even stronger in the case of secondary schools than it is in the case of small colleges. In general, the purpose of secondary-school education in the use of computation should be to enable the student to understand the nature, ease, and power of computation, and to use it in course work in a variety of subjects. This is best done through the use of simplified languages which are not. available on small computers. Any instruction in the nature and organization of computers is best done through special laboratory equipment and experiments, not through operating a computer designed to perform useful calculations. Teaching technical skills in standard programing languages and in computer operation may indeed divert very able and enthusiastic students from an academic career. Such skills, which now have a ready market, may well be rendered obsolete by future rapid developments in hardware and software. However, such training is necessary now, but it is not appropriate as a part of a college preparatory secondary education. APPENDIX G ESTIMATION OF REQUIRED COMPUTER Introduction Estimating the required computer capacity and its cost is difficult because of the complex interrelationship of the needs, the variety of available facilities, and the uncertainty in the possible rate of growth. All of these factors are complicated by the fact that large time-sharing systems for which we have limited experience will be widely available during this period. On the basis of available data and experience we believe that the simple programing languages, convenient terminals, and rapid access of time-sharing systems will lead to a faster growth rate and a more widespread use than with older batch-processing systems. For example, at Dartmouth, within 2 years after installing a time-sharing system, usage grew from essentially zero to the point where more than one-half of the students used the computer each quarter. Further time-sharing systems appear to be an economical means of providing high quality computing service to almost all schools. Purchasing such service is particularly attractive to the large number of schools which do not now have well-trained computer center managers. Consequently, even though many schools may find it feasible and as economical to use another approach; e.g., a very good batch-processing system such as the one at Case Institute of Technology, we decided to base our estimates on the capacity and cost of computing provided by large timeshared centers. We believe that the costs arrived at in this way are close to those for any other efficient and effective means for supplying service; the use of less efficient computers could, of course, lead either to higher cost or inadequate service. Estimation of Needed Capacity The basic unit in the calculation was chosen to be the average number of hours each student is at a console in each week. This figure was estimated as one-half hour per student per week. (Very roughly, this would be equivalent to one-half minute of processing per week on a large batch-processing computer.) It was obtained by estimating that those students making sub 47 |