million. (This increase includes $6 million of the $8 million costs projected for time-sharing systems.)

Combining all these elements, the total level of expenditures which we believe to be attainable and efficiently usable in 1968-69 would be $100 million. Of this, $8 million might be used for large time-sharing systems, $8 million for remote terminals connected to batch-processing systems, and $84 million for computing centers primarily using batch-processing systems. In order to achieve the desired level of computing service by 1971-72, a 60 percent growth rate would be necessary. This would imply a support level of $160 million in 1969-70 and $260 million in 1970-71.

APPENDIX H

THE COMMUNICATION PROBLEM

Everything points to an increasing use of powerful central computers connected to remote consoles of a variety of types. When all consoles are on a single campus, the cost of communication between consoles and computers is small. However, it is highly desirable to provide service to colleges and secondary schools which may be tens or even hundreds of miles from the central computer. In this case the cost of data transmission may be a major obstacle to the educational use of computation.

Various suggestions for cheaper communication have been made. In the case of an integrated system-a State university system, for instance—the State university could operate its own private data transmission network, in which case the educational computing power could be transmitted along those lines. This network might consist of long lines or perhaps microwave relay or coaxial cable. Or, a small piece of an educational television circuit could be used. The data requirements for a teletype machine are small and could be handled by a very small band in the TV circuit.

However, the only communication facilities which are immediately available to interconnect all schools, wherever they may be, are common-carrier communication facilities. It seems possible that common-carrier communication could be provided for substantially less money, both through technical improvements in the use of present facilities, and through providing classes of service more appropriate to present and future needs.

There are present service offerings in which the messages for a group of from 10 to 20 teletype consoles are multiplexed electronically for transmission over a single telephone channel. However, these services are not available in all areas as a standard offering. Work toward improving and reducing the charge for such service might prove highly valuable.

The introduction of new service offerings is more difficult, for it requires both action by the common carriers and approval by State regulating bodies and/or the FCC.

Present services are ill adapted to many computer uses, which call for connections longer than is common in telephone calls, but only a fraction of the time provided by a private line.

A determined study seems called for, involving common carriers, and those knowledgeable in the educational use of computers (the Association for Computing Machinery is a possibility), to seek out ways for meeting the needs of American education.

APPENDIX I

THE GROWTH OF THE COMPUTER INDUSTRY

The purpose of this section is to describe what is known and has been forecast about the needs for computers and for manpower capable of operating and using computers. At the present time, statistics describing the present situation are fragmented and somewhat untrustworthy. To expect precise forecasts in such a rapidly changing field is obviously ridiculous. Nonetheless, the best predictions that can be made might enable academic institutions to make better plans for the future.

There are a number of reasons why statistics gathering and forecasting are difficult. First, the rapid change of computer types and performance characteristics and peripheral equipment makes even the meaning of the number of computers installed difficult to define. Second, as computers get more efficient, new uses and new capabilities will bring new opportunities in fields and areas which cannot be forecast now. Third, the occupational titles and duties vary markedly from one installation to another, and a particular position may be described by over 100 titles.

Despite all these difficulties, a major effort was made by the American Federation of Information Processing Societies (AFIPS) to examine the literature carefully and to examine some of the data and the problems involved in determining manpower and equipment projections. These data have been made available to the Panel and have been the starting point for the summary information described here.

Number of Computers Installed and Projected

The estimated number of computers installed and projected in the AFIPS report is given in figure 1. The rate of increase over the previous 5-year period is also given. "Large scale" computers are defined for the decade of the sixties as those costing $750,000 or more. For the decade of the fifties, the classification was made on the basis of the speed and storage capacity of the computer. Figure 2 shows the increase graphically.

The AFIPS group does not expect the rate of growth of computers in service to increase as rapidly in the future as it has in the past, partly because

the number of computers in service is so large, and partly because the capabilities of computers are increasing so rapidly. It is expected that the total number of computers and related devices will triple between 1965 and 1970, but much of the increase includes expansion of input and output equipment and auxiliary consoles.

The critical point in the estimate of the number of computers needed involves the feasibility and practicality of time sharing. Successful time sharing computers might increase the number of large scale computers even more rapidly.

Value of Installed Computers

Another measure of the increase of computing is the dollar value of the computing installed. Although this measure gives some indication of the man-years of effort that have gone into building computers, the technological revolution which has taken place in computer design and construction does not make these figures an accurate estimate of computing capability. Figure 3, taken from the Rosser report, gives an idea of the increase in computer capability as a function of time. These results should be kept in mind when looking at the increases in value and in the number of computers.

Figure 4 shows the cumulative value of installed computers in the world (leaving out special purpose military computers for which the figures are not available). The table also shows the estimated order of use of computers in foreign installations and also shows the rank in descending order of the countries which use computers the most.

It can be seen that at the present time the United States dominates the world market, but that other nations are moving up, both in use and in production.

Cost of Computing

The cost of computing is much more than the cost of the computer and peripheral equipment (so-called "hardware" cost). Additional costs, usually referred to as software costs, result because programs must be developed for the machines. There are also operational costs, of course.

The U.S. Government has computed the cost for its own installations and determined that software cost is approximately equal to the cost of hardware in its normal use of computers.

The figures given in figure 5 are taken from the AFIPS report. Note that there is a decline in the expected cost of hardware (chiefly because the Government has been moving from a generally leasing policy to a purchasing policy for its computers), but that the software costs are increasing.

Estimates on Manpower Supply and Needs in the Field of Digital Computing

The AFIPS study mentions the extraordinary difficulty of trying to determine the actual supply and the projected needs for computer-related manpower. Job descriptions were not standard, there were no data available with regard to the situation before 1960, and the projections were hard to make because of the lack of knowledge of the type of equipment that would be installed.

Nonetheless, an estimate of the supply and needs was made. The results can be summarized as follows:

1. Hardware systems development. This field attracts bachelors, masters, and doctors, to different levels in the industrial hierarchy. Much of the needed development between now and 1970 is expected to take place in peripheral equipment. Although no clear-cut number of people working in this field was determined, the number was expected to more than keep up with the expected supply of engineers and more than double between 1960 and 1970. The Department of Labor estimates that there were 66,000 production workers in the entire industry in 1964.

2. Systems analysts. This type of employee usually has a bachelor's degree, and occasionally a master's. Because of the difficulties in definition, only estimates of the number of people now in the field and the total expected need, can be obtained. The AFIPS report estimates that there were about 10,000 systems analysts in 1960, 60,000 in 1965, and that perhaps 200,000 will be needed in 1970.

3. Programers. There are widely varying estimates for the number of computer programers expected. Much of the difference in the numbers depends on definitions of programers and on the degree of programing which future computers will require. In any event, most programers have a bachelor's degree training. The estimates are that there were 40,000 programers in 1960, about 60,000 in 1965, and between 200,000 and 650,000 needed in 1970. (It seems unlikely to the Panel that the larger number will in fact be needed.)

4. Computer operators. In general, computer operators have a high school training. In 1965 there were about 43,000 operators. In 1970 the need is estimated at about 80,000. Others estimate this number as high as 130,000.

These estimates show that there will continue to be a need, particularly for college graduates, in the computing field. It does not appear that there will be any enormous shortages, except perhaps in the programing area, if the greatest estimate of need is correct. The salaries have gone up in these occupations, but not spectacularly. It is not certain how many of these employees should be trained in universities and how many will get mostly onthe-job training for routine work.