1. The Working Group was given as terms of reference To assess the probable computer needs, during the next five years, of users in Universities and civil research establishments receiving support from Government funds. At the same time it had to bear in mind the Parliamentary announcement that the review would lead to a new five year programme of procurement which should greatly reinforce the capabilities of Universities and Research Councils and which should start at a rate of £2m a year. We have confined ourselves to the consideration of digital computers, at Universities, Colleges of Advanced Technology, and institutions financed by Research Councils.
2. In making its assessment the Working Group considered that the most advisable way to proceed was, firstly to assess the clear computing needs of Universities and Research Councils over the five year period, and then to assess the likely cost. It could then be seen whether the rate of expenditure envisaged was likely to meet research needs.
3. THE WORKING GROUP RECOMMENDS A RATE OF SPEND FOR NEW OFF-LINE DIGITAL COMPUTER HARDWARE FOR UNIVERSITIES, C.A.T.'S AND RESEARCH COUNCILS FOR THE FIVE YEARS 1965-1970 AS FOLLOWS:
| 1965-6 | 1966-7 | 1967-8 | 1968-9 | 1969-70 | |
|---|---|---|---|---|---|
| £M (ROUNDED) | 1.65 | 4.44 | 4.26 | 5.29 | 5.68 |
4. The total of this is £21.32m, of which £4.75m (22 per cent) is expenditure on American machines. Extra cost will be incurred to the U.G.C. and Research Councils for buildings, operation and maintenance, extra staff, etc., and these are detailed in Chapter VIII. The Working Group considers it essential that the U.G.C. component of these extra charges should be supplied to Universities on an earmarked basis. To enable the programme to start quickly, it is also essential that the costs for 1966-7 should be agreed soon. These figures include a total of £0.45m over the five year period, recommended for a computer in the University of Belfast, which would have to be financed (together with buildings, staff, etc.) through the Government of Northern Ireland.
5. Only a broad estimate has been made of computer requirements for technical colleges, teaching hospitals, etc., and this is detailed in Chapter V.
6. The Working Group has examined the positions in all Universities and in the Research Councils, and has found a most complex existing pattern of computer usage. As might be expected, those departments and establishments associated with the Physical Sciences are the most advanced in their appreciation of computer needs and our assessment of these departments has been the easier for this reason. At the same time, the Working Group is fully aware that other branches of Science and such other departments as Economic Studies and Medicine are rapidly becoming aware of computing needs and the Working Group has tried its best to assess their needs fairly and reasonably.
7. Over the years, informal patterns of co-operation between one computing department and another have been built up and we have found many examples of such interchange. The Working Group recommends that these patterns should be formalised into a regional hierarchical system. Each regional system should be based on one medium or large computer of suitable size, forming a Regional Centre at a University or Research Establishment, with similar smaller computers at individual Universities and Research Establishments, the members of which should, when necessary, have access as of right to their regional central computer. In addition, we believe, the existing ban on departmental machines at universities should be dropped. Such a system, however, only becomes workable if the hierarchical structure allows for compatibility between computers and we have borne this point very much in mind in suggesting specific computer purchases and in making further recommendations on the desirability of good compilers for standard languages, and the need for Research Councils to take the lead in reducing unnecessary duplication of effort by supervising the production of standard major programmes. In particular, we recommend that every computer provided under this programme should wherever possible be capable of operating at least in two major languages, ALGOL and FORTRAN. Where a system has magnetic tape units, one at least should be compatible with international standards; and they should all be provided with card equipment as well as paper tape. Consideration might be given to making the task of ensuring true compatibility with international standards the responsibility of the Ministry of Technology.
8. In making a forward assessment over five years, one is faced with an inevitable problem: the needs of the first couple of years may be seen clearly, but those of subsequent years only with diminishing clarity. What is proposed here is necessarily a rolling five year programme, which will need to be regularly reviewed according to changing needs and the development of new equipment. By agreement with Sir Willis Jackson, the Working Party has undertaken to review all applications for computers at present (June, 1965) before the University Grants Committee. Each has been considered in detail and in many cases the recommendation has differed from the request, in order to fit in with the requirements above for compatibility, and for regional organisation.
9. The outstanding requests of the Research Councils have also been similarly considered. Inevitably, the Science Research Council has had the most ambitious set of requests for computers, and these would be met by our proposals. For the Medical and Agricultural Research Councils, we have, in fact, suggested rather more expenditure than has been specifically asked for, since the Working Group firmly believes that demand for computer facilities will grow more quickly in these subjects than is at present realised. The same attitude has been taken to the Natural Environment Research Council, which has recently come into existence. As a rough guide-line we have suggested that these Councils might aim to spend between 1 and 5 per cent of their total annual budget on computing equipment. We have made no attempt to assess the detailed needs for computers operated on-line in very close association with other equipment. Such machinery should be provided within the normal budget for the research in question.
10. It is possible to assess the needs of both Universities and Research Councils and to decide how these needs may be realistically met, only by considering what are the capabilities of manufacturers, both British and foreign, to meet these requirements within a reasonable time scale.
11. We have regarded the prime purpose of this study as being to assess the computing needs of research institutions and Universities and have regarded the need to stimulate the British computer industry as a secondary, though admittedly very desirable, aim. We therefore allowed ourselves to consider the use of American computers where it was clear that the British manufacturers could not reasonably be expected to supply adequate hardware within the time scale.
12. By arrangement with the Ministry of Technology, the Working Group met the representatives of three main British manufacturers, and also those of two major American firms, I.B.M. and C.D.C., in order to discuss, without commitment, the likelihood of demands being met. As a result of these conversations, it became clear that British manufacturers could meet all the likely requirements of research organisations for medium and small computers (approximately of the size of an upgraded KDF 9 and downwards, see main text). The Working Group firmly believes that it is necessary to acquire much larger computers within the next five years. Some forms of calculation are possible only on such machines; they represent the most economic way of acquiring a large amount of computing capability and they seem the only way in which the requirements of the more active of the regions can be met.
13. The Working Group recommends the acquisition of several large American computers; the precise number is detailed below. It is suggested that most of these American computers be hired rather than purchased. We believe that by 1970 there will be a requirement for some 10-15 large machines in this country (including the requirement of certain research organisations outside the scope of our terms of reference). Such machines should be modular in design and of size between 3 and 100 times that of an I.B.M. 7090.
14. In one case there is an additional reason why we are anxious to recommend purchase of an American machine. This is at Edinburgh where we recommend the immediate development for the joint use of Universities and Research Councils, in conjunction with a suitable manufacturer, of a pilot scheme for a multiple access system. We regard such a development as a most important addition to British computing experience.
15. The details of recommended computer expenditure and of how we arrived at them have been given in the text of this report. The cost of outright purchase of all the machines for Universities, C.A.T.'s and Research Councils for which we see a clear need within the five year period together with building and operational charges, and an allowance of 5 per cent for extra peripherals arising as contingencies, would be £35m, of which £26m is strictly the cost of the machines, £10m of this representing the cost of American machines for which we see no reasonable British equivalent within the time scale. Of this sum £0.6m (£0.45m hardware) would be the responsibility of the Government of Northern Ireland.
16. Any attempt to provide less than this computing facility would represent a serious handicap to the progress of research and higher education in this country. We have studied the way in which this expenditure will add to the volume of computer facilities year by year, and found that it represents an average growth factor of about l.8 each year. This we regard as not unreasonable and as representing the normal way in which experience shows that a demand for computing facilities grows. As mentioned in the main text, the cost of supplying such a growth factor will eventually not increase by anything like a factor of 1.8 a year, because of the way in which large computing facilities become markedly cheaper with time.
17. Although these figures represent the cost of purchasing outright the computing facilities which the Working Group considers are needed, it is not recommended that all these sums should be so spent. It is believed that the large American computers should be hired rather than purchased, and replaced with large British computers, if and when these become available. In these circumstances, the total cost of our recommendations in the five year period is £30m (including buildings etc.) of which £21.3m represents the cost of hardware (22 per cent of this being the cost of hiring American computers). These are the figures given on the front page. For the reasons given, we regard the expenditure of any sum of money significantly less than this as falling short of meeting the needs of these Institutions. Further sums, very roughly of the order of £2m, are needed for Technical Colleges and teaching hospitals, and these sums should be met from other sources.
18. We are aware that the growth rates recommended for computer costs are greater than those envisaged for scientific research expenditure in general. We regard this as an inevitable consequence of the need of British scientific research to make up lost headway in acquiring computing facilities, and we believe too that, by spending money at this rate, we will help to stimulate the British computer industry to regain its proper place as a viable commercial venture.
19. THE WORKING GROUP THEREFORE RECOMMENDS:
