The Committee approved the Minutes of the Meeting held on 8th October 1965. It was agreed that matters arising from the Minutes would be dealt with during the discussion on the various papers before the Committee.
Dr Howlett presented his report on the Laboratory.
2.1 The Machine: The Committee noted that the machine was about half way through the final three months test period and that while it seemed probable that the conditions for percentage good time would be met it was possible that the fault rate per shift week might slightly exceed the maximum allowed.
It was agreed that in the event of the final acceptance conditions not being met the Chairman should be consulted (after any necessary preliminary discussions with ICT Limited. The Chairman would at his discretion agree on a course of action or consult further with the Committee.
2.2 Compilers: The Committee noted that the use of Fortran was growing at the expense of Algol (of the total work some 60% was Fortran as opposed to 8% Algol). It was, however, valuable to have the various dialects of Algol available. There were no firm plans for implementing PL/1, but the possibility of accepting some subset was being explored.
2.3 Work Load: There had been no major changes in the work load. Dr Pickavance said that the Rutherford Laboratory was very pleased with the service they received. Some of the major university users would have machines of their own in operation during the next six months, and any consequent changes in distribution of university work would be a valuable indication of the future pattern of the work load.
The Committee noted that the Laboratory was taking 10-15% of the available time of which perhaps 3% represented internal research and the remainder systems work. This was considered reasonable, but it was agreed that the Laboratory would not make any substantial increase in the time allocated to internal research without the Committee's agreement.
2.4 Postal Survey: The second survey sample showed 10% of jobs delayed by four days or more. This was mainly a consequence of the priority given to engineering work in the period, and the effect of week-ends and of long jobs.
It was suggested that special arrangements might be made with the Post Office giving a further marginal improvement in turn-round.
3.1 The Chairman said that the proposals in the Report had been given general Government approval although in terms which might present certain financial difficulties for universities (for example, over extra building and some recurrent costs). It was to be hoped that a formal announcement on the Computer Board would not be delayed unduly as the UGC's interim Committee inevitably felt left their freedom of action constrained.
There were two particular points which called for comment. Firstly, it had been suggested in the Press that the Report recommended purchasing 10-15 large machines from the USA in the second five-year period. This was not correct; the Report estimated that this number of machines would be required but it did not suggest they must be American, rather this estimate was a challenge to UK production. Secondly, paragraph 325 of the Report suggested that an indefinite exponential increase in computing demand could only be controlled if users were faced with a real economic choice. It was not suggested that there should be any immediate change in, for example, the Laboratory's policy of a free service t universities, but once the present grievous shortage of computing power in the country had been overcome demand should be tested by money criteria.
Finally, the whole spirit of the Report was that of a first attempt at a rolling five year plan. The field must be reviewed approximately yearly, a task which would fall to the Computer Board.
3.2 Dr Howlett drew the Committee's attention to a similar survey conducted in America entitled Digital Computer Needs in Universities and Colleges published by the National research council's National Academy of Sciences. Copies of this report were being obtained and would be circulated to the Committee.
3.3 In discussion it was agreed that some effective method of controlling and rationalising demand would be required. In general form this would probably entail a gross allocation between universities within which each university would have to decide its own priorities. Money might be one way in which choices could be evaluated and this might prove feasible with the growing university interest in techniques for costing experimental work. It was recognised, however, that without suitable precautions a bias might be introduced against certain users: for example, the arts and social sciences, where the scale of finance was very different to that in the physical sciences or against relatively junior research workers trying to break new ground. These dangers, avoided under a time rationing system, would need to be weighed carefully against the advantages introduced by genuine economic choice.
3.4 Professor Flowers agreed that if the Report had been written six months later there would have been a somewhat greater emphasis on Multiple Access techniques. This was one example of the very rapid changes taking place in the computer field.
4.1 The Chairman observed that his Report, while not attempting to deal with the long term future of the Laboratory, had recommended the continual purchase of substantial items of equipment at a rate of about £200,000 per annum and had included a purely notional sum of £2M in the fifth year in case the Science Research Council should find it necessary to replace Atlas with a more powerful machine. He then invited the Chairman of the University Science and Technology Board to explain how the Board regarded future developments at the present time.
4.2 Professor Sir Ewart Jones thanked the Committee for their invitation to the meeting. The Board recognised the very important role which the Laboratory would play in computing support for universities in the next two years. Further, the Board wished to do all it could to ensure the present machine was used to its full capacity. To this end approval in principle had been given for the purchase of a disc store and more recently, in considering the forward look, the Board had given high priority to the magnetic card file alongside the other comparable proposals for expenditure. The Board had found its position more difficult in regard to the major question of a replacement costing some £2-3M. Here there were two other major proposals of the same order of cost, a High Magnetic Fields Laboratory (long agreed in principle), and a High Flux Neutron Beam Reactor. Provisionally, and lacking the advice of the Committee, the Board inclined to give higher priority to the last two proposals since these provided facilities unique in the country.
The fate of all three major proposals depended on the rates of growth agreed for the Council and on the Council's allocation between Divisions. The most the Board could expect was a 15% increase in 1967/8 reducing to 12% by 1970/1. With this it would be possible to finance the High Magnetic Fields Laboratory and the High Flux Neutron Beam Reactor, but not a major replacement for Atlas. There were, however, uncertainties in the situation. Firstly, the two schemes were to be in collaboration with the Atomic Energy Authority and it was not yet certain they would proceed. Secondly, certain assumptions had been made regarding the rate of growth of demand for Research Grants; these were difficult to forecast and if the 12% rate of increase assumed proved to be too high the situation would be eased. Firstly, the general timing of major proposals produced a peak demand for money in 1970/1 and 1971/2; if the replacement for Atlas could be postponed finance might be easier.
4.3 The Chairman thanked Professor Jones for his statement and for the helpful attitude taken by the Board. Dr Howlett then introduced his paper on the future development of the Laboratory, stressing in particular:
4.4 The Committee considered that the principle factors on which a decision should be based were:
In view of these factors the Committee recommended:
In making these recommendations the Committee recognised that in a rapidly changing technological field all three recommendations would naturally be subject to continuing review and that the first did not exclude the possibility of eventual organisational changes including resiting the Laboratory, say at Oxford, in some years' time.
The Committee took note of the minutes of the twelfth and thirteenth meetings of the Atlas Users' Committee.
Mr Jolliffe explained that although the finance timetable had entailed the preparation of the forward look for consideration by the Council on the 14th April, any observation of the Committee could be reported through the Chairman and taken into consideration in the discussions with the Department of education and Science.
The Committee noted that the effect of their recommendations under Minute 4 above would be to increase the capital estimates before the question of replacement for Atlas was taken into account by £200,000 a year in 1968/9 and 1969/70. This was probably not a significant change in the context of the Council's estimates as a whole and their probable margin of error. The effect of the Committee's recommendations on a major replacement would make it unlikely that there would be any expenditure in 1968/9 or 1969/70 but substantial sums would be required (of the order of £1M a year) in 1970/1 and 1971/2. This would still leave the heavy expenditure in a year where the pattern already peaked, but in view of all the uncertainties surrounding forecasting of major capital schemes at the end of the five year period the problem might prove more apparent than real.
Dr Howlett said that since his paper had been prepared ICT had reconsidered the relative merits of the Bryant and Data Products discs and now inclined to the latter. It was probable that the Data Products disc would be chosen for the Manchester Atlas.
The Atlas Laboratory would be responsible for the software and ICT for the engineering. With a disc store the number of tape decks could be reduced by about four. There would be no serious loss of machine time while the disc was fitted; the drum coordinator modifications would be carried out at week-ends and software development piecemeal.
The Committee recommended the purchase of a 25 million word disc store with dual read-write equipment at a cost not exceeding £233,000 (including duty if payable). It was further agreed that the Chairman be authorised to agree the make of disc and any other technical details on the Committee's behalf.
The Committee noted that consideration was being given to the acquisition of a microfilm recorder. The choice seemed to lie between a Stromberg Carlson 4020 (purchase price £86,000) and a similar recently developed device of Benson-Lehner (purchase price £53,000). Formal proposals would be put to the Committee at its next meeting.
The Committee noted that technical discussions on a magnetic card file were taking place with ICT and that the cost would probably be of the order of £100,000. Formal proposals would be put to the next meeting.
It was agreed that in future meetings should be arranged to fall before meetings of the University Science and technology Board. Unless any urgent business emerged the next meeting would be held in December 1966 and subsequent meetings in July and December of each year.
