Apologies were made for the fact that Mr B R Taylor had been incorrectly represented as Dr R B Taylor on page 1. The minutes of the last meeting were then approved.
There were no matters arising from the minutes.
This item was taken first because Mr Jolliffe, Dr Pickavance and Mr Walker had to leave early.
Introducing the paper Dr Howlett said that his proposal was essentially a gradual change in priorities. at present the Laboratory's practice was to take work from all universities and in virtually all fields and to give fair shares to all users and no preferences. In the regime he was proposing work in certain selected fields would be given preference by being guaranteed time, possibly in large amounts. Any time remaining after the guarantees had been met could be used for general service work. selection of fields to be given preference would be made by taking the best scientific advice available in the country, as was in fact available to the SRC through its Boards and Committees. as was stated in the paper, the change would necessarily be gradual and might take two to three years to complete.
Mr Jolliffe said that the Council itself was developing a policy of selectivity in its support for science and that the proposals of the paper were consistent with that policy.
Mr Hailstone said that his impression was that the total computing capacity in universities was still much below the demand and would remain so even when the 1906A were installed; and that university people were looking to Chilton to make up the deficit and to help them in emergencies or other times of special difficulty, such as when moving their own machine to a new building. He said that the demand was spread very widely over universities and over subjects, and Dr Churchhouse said that only about 20% of Atlas's time was taken up by users of the major packages.
Discussing machine purchase in general, the Committee noted that there had been a strong desire, but no directive to buy British. The Ministry of Technology, with its responsibilities for the industry, would naturally try to persuade potential buyers to do so but would not oppose a sound case - as was shown by the purchase of a big CDC 6600 system for London. Thus the decision on the Laboratory's future did not have to assume that only a British machine could be bought. However, a possible policy could be that the Laboratory would always take the biggest available British computer (which would now mean an ICL machine) and feed back its experience to ICL. it was acknowledged that the Laboratory could not be expected to write the operating system or any considerable part of it, because this had now become a very large undertaking and one into which all manufacturers put very large resources.
The Committee discussed the possibility of the Laboratory selecting its work-load by size of job, or by type, meaning by the latter the kind of demand the job made on the machine. the former seemed not to be a good criterion, both because it gave no automatic guarantee of value of the work and because it was subjective - a project which was large to a potential user who had, say, an ICL 4130 in his own university could be very small as an Atlas or 1906A job, and smaller still for a CDC 6600. Also many small development runs were needed to get a big program going, so the problem of handling large numbers of short jobs remained. the latter also seemed too indefinite; one could envisage projects of extremely special type which would demand very large and specialized machines, but the general feeling in the Committee was that no case had been made for such specialization, and that the modern computer was flexible enough to deal with at least the great majority of types of problems.
it was recognised that the decision to select certain fields for support implied a withdrawal of support from others, and that this could produce pressures for support for those fields by other means.
The possibility of the Laboratory having a very large machine with data links to terminals (either simple input/output devices or small computers with their own storage) in many or all universities was considered. The Committee expressed its doubts on the soundness of such a policy: transmission costs would be high and unless most of the terminals were very lightly used, and therefore an uneconomic capital investment, the service would be degraded by the need to serve so many links. The Committee thought it likely that a computer network would be developed in the future and that the Laboratory could become one of the nodes. It recognised that there would be problems in linking machines of different makes.
The Committee agreed on these points:
Summing up, the Chairman said that there was sympathy for the proposals of the paper, and that it would be possible to experiment with a policy of selectiveness without committing the future irrevocably. The Committee would need a more precise proposal before it could make a decision and he invited Dr Howlett to formulate one for discussion at the next meeting.
The Committee noted the tables and the general report. Dr Howlett added two points:
The Committee agreed that this was a situation in which the Laboratory could do nothing, because it had to rely on AEA Production and Works Division for estimation of building costs and for scrutiny of tenders. Mr Laver said that air-conditioning plants had always to be studied with great care and that the experience in the Post Office was that this gave more trouble than any other part of a computer installation. thus whilst he regretted the need to ask for more money for the building, he felt it would be foolish to risk any under-provision in this part.
The Committee accepted the arguments of the paper for an experimental study of the use of remote terminals with the 1906A. they felt, however, that because of the high cost of the card reader/lineprinter terminals, the uncertainty of the load which the 1906A could bear and the absence of a policy for deciding in which universities such terminals should be located, the experiment should start with two links and not the four asked for. They therefore approved the proposals of the paper with the change that two only of the Type 7020 terminals should be bought.
Dr Howlett reported that an unexpected outcome of the agreement between the NCC and the SRC for marketing ASCOP was that any university wishing to have a copy of the program would have to pay £2,000 to the NCC. This apparently was in accordance with the Council's finance policy, and there appeared to be no escape. The Council could supply a copy or the money to buy one to a university which had an SRC grant which required the use of the program on its own machine. Mr Smith confirmed that this was indeed the position, and that it had been reached after discussion at a high level. The Committee were exceedingly distressed to learn of this and asked to have put on record their very great displeasure that such a restrictive agreement had been made. They felt that an inevitable effect would be to curtail the use of ASCOP in the university world. They recommended that, before any other programs produced in the Laboratory were handed over to NCC or any other body for marketing, conditions should be included in the agreement to prevent the same thing happening again.