Any comments on the decision by Council to establish a SERC Common Base Policy probably needs to answer the following questions:
While these questions should be answered by somebody not involved in the project, some suggested responses might be as given below. One should judge the project on the knowledge available at the time rather than assuming 20-20 hindsight.
Need for a Common Base: the major reasons given for the need for a Common Base were that a large number of systems would appear and without some kind of standardisation, scarce resources would be diluted in supporting too many systems.
Over the last few years, the number of systems has increased as anticipated. Probably the growth was even greater than anticipated. When the assessment of systems on the market took place nearly two years ago, well over 100 companies had products which needed to be considered.
The tremendous impact of the Apple Macintosh and similar systems has meant that the mode of working predicted as early as 1979 has definitely come to pass.
Choosing the PERQ: assuming the decision was to be made in the period 1980-81, there were only two possibilities - Apollo and Three Rivers. Apollo have maintained their company stance of providing a complete system where interfacing other systems is difficult and costly. In a university research environment, this is not the best choice and remains so today. Apollo improved their product quickly over the next two years. If the decision had been made a year later it might have been different but not obviously so.
Given ICL had put their full weight behind the PERQ project, it clearly had the potential to be a major force. ICL, despite many problems, is still probably the third largest supplier of single user systems in the world today.
The PERQ was seen to have advantages in its bit slice architecture and quality graphics. To some extent, the need to go this way for high quality products has been vindicated by the competitors where Apollo have moved to this architecture at the top end. SUN also have plans in this direction.
The arrival of the 68020 systems means that the current PERQ system is obsolete. However, the plans at the time were to have the new PERQs in production by 1984. There is no reason why that could not have been achieved.
Cambridge Ring: at the time the decision was made, there were many different variants of ethernet and the JNT had established a university standard in the UK around Cambridge Ring. The Engineering Board had invested heavily in Cambridge Rings. It was the right decision at the time.
The decision by the JNT and UK industry not to press ahead with international standardisation of the Cambridge Ring until it was too late was a major failure. That decision was outside the control of this activity.
UK Company Involvement: SERC has consistently attempted to assist UK industry where possible. A number of Council members have been very active in this area. SERC receives pressure from other government departments to move in this direction. It was clearly in line with Council policy at that time and now.
Choice of ICL: the three companies that were approached were Logica, GEC and ICL. All three had the capability to manufacture PERQs in the UK. GEC had their own plans and were not interested. Logica showed interest but eventually did not have the commitment. ICL's manufacturing capability is clearly the best of the three and the company has the ability and know-how to manufacture competitively in a volume market.
Of the companies not approached, Racal and Plessey perhaps should have been tried. Recently, Racal has shown an interest in this area.
Timing: the decision would have been different later as there would have been more choice. If the decision had been a year later both Apollo and SUN would have been alternatives. The early SUN products were inferior to the PERQ and the reasons against Apollo would have remained. Consequently, if the delay had been only a year, the decision would not have been different.
If the delay had been three years after the appearance of products from IBM, HP, DEC and many smaller companies, the position would have been similar to when the assessment was carried out for a hardware system to join the PERQ in the Common Base. After benchmarking a range of systems the outcome was that none of the systems on the market were up to the standard required of a system to enhance the Common Base.
You would probably have needed to wait until 1984 before a significantly different decision would have been made and that would probably have been to base the Common Base on the SUN workstation.
In the period 1980-1984, experience gained on single user systems would either have been fragmented or lost. It is unclear whether waiting would have been an advantage. It is probably one of the weakest areas in terms of justifying the Common Base. However, most of the counter examples are dependent on hindsight of what happened to ICL and its competitors rather than what might have been.
UNIX: the decision to go for UNIX as the operating system has been totally vindicated by the market place. Nearly every single user system on the market of this power runs UNIX as the operating system. Even Apollo has been forced to mount it as an alternative to their proprietary system.
EB local initiative: to get ICL to commit required a level of commitment in SERC which was difficult for the Engineering Board to underwrite. In retrospect, the Engineering Board has committed the level of finance required and a local EB initiative would have been a better policy.
The decision by Council to go for a more broadly based approach was probably correct in theory. All Boards will eventually use these systems. However, the lack of commitment by other Boards and the decision not to back the project at the level required seriously weakened the programme. If the Engineering Board had remained responsible for the project, it would almost certainly have received a more sympathetic hearing and a firmer level of commitment.
My own view is that it was a major mistake for Council to go for a SERC-wide Common Base without ensuring the necessary level of commitment from the other three Boards.
Facts to Council: the relevant papers are provided on this site. RAL believes that they give an honest view of the situation given the degree of commercial confidentiality that was essential in a joint project of this type.
Funding: the need for 16.5 MYs of effort in the first year to manage the project and implement UNIX was probably a correct assumption. The amount of effort actually used can be checked.
To a large extent, a decision in this area needs to be left until the next part of the PERQ history is described.
In summary, there is good evidence to suggest that a Common Base Strategy for Single User Systems was timely and correct at the time it was initiated. ICL was the logical choice of UK company to be involved. The PERQ was the best system on the market at that time. An attempt to have a Council-wide policy was sensible of Council. Perhaps, the dangers of such an approach, knowing the four Boards, should have been rea1ised and a greater measure of commitment established. The programme clearly had strong support from the Engineering Board and, if it had stayed as an Engineering Board project, things would have been different.
