277. Having considered the special problems associated with Universities and Research Councils, we wish now to comment upon the present overall situation in Britain relative to that in the United States, taking into account the whole public sector. This is useful in that it helps to establish a scale against which our proposals can be measured. It will become apparent from this study that although we lag relative to the United States even in the total number of computers available to us, our most serious deficiency is in large computers which alone can make possible a whole range of modern scientific and technological ventures. Our proposals, if accepted, will go a long way towards rectifying this very serious situation. We also briefly examine in this chapter the position of the U.K. computer industry relative to that in the States, insofar as is relevant, directly or indirectly, to our enquiries.
278. To fix an absolute scale, the assumption might be made that Britain ought not to spend a smaller sum on computing, expressed as a fraction of the Gross National Product, than does the U.S.A. Such a condition will be termed parity; it means in practice about ⅛ of the U.S. expenditure. On this basis, less computing power will still be available for each British computer user than is the case in the U.S.A.- by a factor of order 2 - but this disadvantage might well be removed by a strong drive for increased efficiency in the use of computers. Such a drive could be supported by the Government, and would not itself be expensive. A case could even be made for spending initially a larger fraction of the G.N.P. than does the U.S.A. in order to improve our relative industrial position, since the fraction is in any case extremely small. (In the U.S.A. the accumulated total spent on hardware so far amounts to only 0.6 per cent of the yearly G.N.P.).
279. The total numbers of computers of all types (research, general governmental and commercial users) in the U.S.A. is estimated (July 1964) at about 15,000*, increasing by about 25 per cent per annum, with a value of about $4,000 m. (The computing power as we have seen increases much more rapidly, since faster machines are continually being introduced). On this basis, the number installed in Britain for all purposes ought by now to be 2,125, of value about £180 m. In fact it is about 1,000. The total number installed is therefore only half of what it should be to achieve parity. In terms of actual computing power, the situation is more serious, as will now be shown.
* Numbers of U.S. computers in this chapter are taken from International Business Automation, October-November 1964, and numbers of British computers from Computer Survey, March 1965. Financial values represent approximate estimates only.
280. A detailed analysis of U.S. machines at July, 1964 has been made. A remarkable feature is the number of installations costing more than $750,000, amounting to 1,510. On this basis, to achieve parity the number in Britain should by now be about 240; in fact it is about 30. Since machines in this price range are much more powerful than the small business machines which account for most of the total numbers, the shortfall in national computing power is very substantial. The cost of the missing machines is also large; even if we adopt the minimum of the range it exceeds £50 m. It is in this light that the present Government proposal to spend £2m in the first year of the five year programme on computers for research should be viewed.
281. Of the 30 larger machines in Britain mentioned above more than half are IBM machines. Tables 2 and 3 indicate total sales by U.S. and British companies. The basis is not quite the same in each case, since Table 2 refers to July 1964 and only to machines still in production, while Table 3 refers to the accumulated total of all machines at March, 1965. As a rough estimate, figures in Table 2 should be increased by 1.4 to get a true comparison. The percentages have been adjusted to allow for this. The Tables indicate how dominant is the position occupied by IBM with 68.5 per cent of the world market. By contrast, the largest British firm (ICT) has only 1.9 per cent, while the next largest U.S. firm has 5 per cent. (These percentages represent numbers not value or power).
282. If sales of a given computer are too few, hardware and software development costs are spread over too small a market, and the company will not be viable. To take an example, the total sum spent on University KDF 9 computers so far is about equal to the cost of one IBM 7090; yet IBM have sold over 200 machines of this kind alone and were able to make use of earlier development work carried out on the 704 and 709. Other KDF 9 computers probably increase the total by a factor 4; out of this English-Electric-Leo-Marconi have had to support two entirely different software systems.
283. Although not within our terms of reference, we consider it appropriate to record our view that it might be undesirable if the whole economy came to rely on the success of one or two small home manufacturers. Such a situation might also discourage increased efficiency at home through lack of competition, and lead to British techniques becoming out of step with world standards. At the same time, computers do appear to form an ideal product on which to concentrate. The cost of raw materials is small, and a number of significant developments require only common sense, intelligence and organisational skill of a kind that British designers have demonstrated in the past. This is particularly true for software. In addition, most competing countries are now out of the market.
284. If consumption had reached parity, each of three British manufacturers would by now have supplied about 700 machines, assuming that exports and imports approximately balance. This would correspond at July 1964 to a figure of 500-600 in Table 2, and it will be seen that this is quite comparable with the larger U.S. manufacturers, excluding IBM. Therefore, in our view there would be a good chance of these companies being viable, provided they did not each try to cover the full range of machines independently.
285. The Government is invited to consider the existing computer shortage and realise its responsibility for removing it as rapidly as possible, since it influences or controls the spending of such a large fraction of the potential users, namely Universities, Technical colleges, Schools, Research Councils, Defence, A.E.A., Government departments, G.P.O., Local Authorities, and Nationalised Industries>
286. The cheapest way of meeting the shortage of national computer power would undoubtedly be to take advantage of recent American developments by installing a relatively small number of the largest available machines, for which the present contenders are the CDC 6800, probably available in 1967 and the IBM 360/92 somewhat further off. Both of these could be approached in the future by upgrading smaller machines in the ranges available now. Assuming very tentatively that the power of these large machines, which are not yet available, might be equivalent to that of 50 IBM 7090 machines, and that the missing 210 machines costing $750,000 and upwards each have an average power equivalent to half an IBM 7090, only two large machines would be required at this moment in time to overcome the shortage; allowing for an annual growth rate in computing power of 1.7, this number would become 6 in two year's time, at a cost of around £12m-£15m. Again, this merely emphasises the high effectiveness/cost ratio of the large machines.
287. To be set against a solution based solely on large machines must be the equally powerful argument that computing at a distance for the majority of purposes is time-consuming, frustrating, and detrimental to the development of the art. Although the total cost is greater we have tried to seek a compromise solution, at any rate for the Universities and Research Councils, based on regional hierarchies which combine the provision of smaller local computers for normal purposes with the availability of a few larger machines for exceptional purposes. This has the additional merit of enabling British manufacturers to play their proper part, and to draw to their attention the growing demand for larger machines, at present available only from American manufacturers.
288. We have recommended two very large American machines for the Universities and two capable of upgrading to the same standard for the S.R.C. over the five year period. These alone, even though they are to be supplemented by a much larger number of small and medium machines, will not be enough to overcome the shortage of overall national computing power as determined by the considerations of parity for this country as a whole. We therefore recommend that the Government should consider very carefully whether it can afford not to provide a few more very large machines at establishments outside our immediate concern in this Report.
289. It is important that (like the U.S. Government) the British Government should take a strong positive initiative in furthering all imaginative and realistic applications of computing, instead of simply assessing applications for grants as they arise. This would involve detailed knowledge and access to information, surveys of the equivalent U.S. position over the whole public sector, search for possible new techniques, and formulation of a coherent plan continually updated as the situation changes. In particular, an investigation should be made to find how computers can best be introduced into schools and industries, possibly by using mobile equipment; a pilot experiment could be initiated very quickly.
Taken from International Business Automation for Oct.-Nov. 1964 except for last column: (see note (2))
| Manufacturer | Over $750,000 |
$100,000 to $749,000 |
Under $100,000 |
Total | % World Total |
|---|---|---|---|---|---|
| Adressograph-multigraph | 10 | 10 | 0.06 | ||
| Adv. Scientific | 5 | 20 | 25 | 0.16 | |
| Burroughs | 23 | 260 | 283 | 1.70 | |
| Clary | 180 | 180 | 1.10 | ||
| Control Data | 80 | 434 | 514 | 3.20 | |
| DPD | 26 | 26 | 0.16 | ||
| Friden | 137 | 137 | 0.85 | ||
| G.E | 58 | 112 | 170 | 1.10 | |
| General Precision | 791 | 791 | 4.90 | ||
| Honeywell | 63 | 104 | 167 | 1.00 | |
| I.B.M. | 891 | 8,503 | 1,631 | 11,025 | 68.50 |
| I.T.T. | 9 | 9 | 0.06 | ||
| Monroe | 396 | 396 | 2.50 | ||
| N.C.R. | 27 | 187 | 588 | 802 | 5.00 |
| Packard-Bell | 4 | 178 | 182 | 1.10 | |
| Philco | 24 | 15 | 39 | 0.02 | |
| R.C.A. | 4 | 568 | 572 | 3.60 | |
| S.D.S. | 2 | 128 | 130 | 0.81 | |
| Teleregister | 14 | 10 | 24 | 0.15 | |
| Univac | 127 | 127 | 0.79 | ||
| Total of Units in Production | 1,320 | 10,214 | 4,075 | 15,609 | |
| No longer in Production | 386 | 2,795 | 635 | 3,816 | |
| Total | 1,706 | 13,009 | 4,710 | 19,425 |
1. A machine such as the IBM 704 is out of production, and not included in the body of the table.
2. World total is U.S. units at July 1964, plus an approximate allowance for British manufactured machines (other countries ignored)
3. Percentage is by number of machines, not value or power.
TABLE 2 ESTIMATE OF NUMBERS OF MACHINES SUPPLIED BY U.S. MANUFACTURERS TO ALL COUNTRIES, UP TO JULY 1964
| Manufacturer | Number | % World total |
|---|---|---|
| A.E.I. | 10 | 0.04 |
| English Electric-Leo-Marconi | 114 | 0.50 |
| Ferranti | 11 | 0.05 |
| I.C.T. | 439 | 1.90 |
| NCR/Elliott | 318 | 1.40 |
| Total | 892 | 3.90 |
TABLE 3 ESTIMATE OF NUMBERS OF MACHINES SUPPLIED BY BRITISH MANUFACTURERS TO ALL COUNTRIES UP TO MARCH 1965
1. World total is British + U.S. total at July 1964 + allowance of 2640 for estimated increase in U.S. total to March 1965, making 22,957 in all (other countries ignored).
2. Percentage is by number of machines, not value or power.
3. Numbers of British built computers taken from Computer Survey for March 1965.
