Flowers Report

Jump To Main Content

Jump Over Banner

Home

ACLLiteratureManualsAtlas Autocode

Jump Over Left Menu

Chapter IX Postscript

321. As was explained in Chapter I, the Working Group, for reasons of time, has had to confine itself mainly to the needs of the Universities and Research Councils. The needs of other Government establishments and of industry we have hardly touched upon; nor have we dealt, except briefly, with the Technical Colleges, or the schools, or hospitals other than those associated closely with Universities. These matters, we feel, are best handled by separate exercises under the Departments primarily concerned. Nor should it be forgotten that many of the requirements of defence will be similar to those we have considered.

322. We have proposed a system of regional hierarchies of computers serving the Universities and Research Councils. We suggest very tentatively that in some way these regional arrangements might in the future be correlated in some way with the Economic Planning Regions to facilitate a wider application of computer technology than we have been primarily concerned with. As long as sufficient computer power is provided, there would be many advantages in industry and local government contributing to, and making use of, the regional hierarchies. However, a much more detailed study would be required before it could be seen how this could be achieved in detail.

323. We wish to turn now to the longer term outlook. The needs for small and medium sized machines (in the sense used in Chapter IV) will continue to grow, although in the future many of them may be operated as satellites of a larger machine within a single University or a large research laboratory. We are more concerned with the larger computers not at present available from British industry, and our work has made clear the need for a modular series covering the power range from about 3 to 100 times that of the IBM 7090. Beyond 1970 the upgraded University KDF 9 machines, each equivalent to a little more than the 7090, will be thoroughly overloaded by that time and will need phased replacement. At least two other large regional machines will be needed over the following five years. In addition, the Atomic Energy Authority may require at least one machine in this class. If the 300 GeV European accelerator comes into being, there will certainly be a need for several very large machines associated with that project. Thus in the period 1970-75, we can foresee a minimum requirement for 10 to 15 machines in this range, some of which should be very large indeed and many operating on multiple-access lines.

324. Our enquiries have brought to the fore another matter of interest. At several points in our Report, we have emphasised that the physical sciences, which have hitherto provided the main driving force for large scale computation, are likely to be overtaken in the demands they make by other disciplines, some of which have radically different technical requirements. The processing of very large amounts of data for information retrieval, for example, requires, vast amounts of quick random access storage rather than further sophistication of arithmetical units. We urge industry to bear this in mind in their development plans.

325. The demand for computer time is growing very rapidly. At some point in time, however, computer provision for research must be absorbed within the overall growth rates approved for science although we hope that this is not done until the leeway is made up. But whatever provision is made, it will undoubtedly saturate rapidly, so generating further demand, unless each and every user is posed with an economic choice. At present, computer time is usually provided free to the individual user, at least in the Universities. He does not have to pay for it out of his own departmental research budget. It might often be cheaper to do an experiment or otherwise alter the programme of work than to compute, if the cost of computation were set against individual research budgets. Some difficulties would undoubtedly arise in asking each user to pay for computer time in cash; but this is the only way we can see for introducing some negative feedback into an otherwise explosive situation.

326. Finally, we wish to draw attention to a matter not strictly within our terms of reference. The Working Group has been left in no doubt that its labours have been worthwhile and have been welcomed by those concerned. We believe that we should be succeeded by a suitably constituted permanent body, responsible to the Department of Education and Science, with a continuing responsibility for co-ordinating the provision of computers to Universities and Research Councils on a rolling five-year basis, and for the continuing reformulation of policy in a field in which technological change is very rapid. At present the U.G.C. is responsible for the provision of general purpose computing facilities to Universities, the Research Councils for the provision of computers to their respective organisations and for special purposes at Universities; but the hierarchical organisation we have recommended, based primarily upon geography, is one which pays rather secondary attention to the distinction between University and Research Council laboratory.

327. The agency could have as parent body the U.G.C., one of the Research Councils, or the Department itself. That must remain a matter for more detailed discussion. However, we would strongly recommend that its terms of reference should be such that, within an approved policy and an approved budget, it can operate independently and quickly. Such terms of reference might be:-

  1. to formulate and submit to the parent body a national policy for the provision of computers for research, together with forecasts of expenditure.
  2. to execute on behalf of the parent body, and within approved estimates, the approved policy.