History of Computing in France: A Brief Sketch

Those in charge of the Plan Calcul regarded informatics as the driving force for innovation in the semiconductor industry: this indeed was one of the main reasons for launching it (Allиgre 1969). Unfortunately, the informatics industry which they set up did not constitute, for Sescosem, a large enough outlet to justify a decisive reorientation towards computer components: the 246 systems installed by CII between 1967 and 1970, a high proportion of which were SDS "minis", were not enough to persuade the French electronics industrialists to make the massive investments that would be needed to put digital integrated circuits on what was a hyper-competitive market. Such circuits represented less that 0.2% by volume of Sescosem's production in 1969. In its forecasts for 1974-1975, SESCOSEM did not hope to sell more than 20 MF of ICs for computers (8% of its expected revenues), out of which CII accounted for one quarter. The CII engineers modified the SDS machines by replacing the original DTL circuits by TTL; these gave higher performance, but most had to be bought from the leader in this technology, Texas Instruments.

Since it is difficult to find any written information about the semiconductors purchased by computer companies, I opened a few CII machines and examined their components. Most of the integrated circuits came from Texas Instruments. For instance, I dived into the entrails of an IRIS 42 and an IRIS 50 (medium-sized business machines, built around 1970) and a Mitra 15 of 1973. In the IRIS, about 90% had the TI logo alongside the CII attribution mark; COSEM (CSF) produced the remaining 10%, and most of the discrete components (diodes and transistors). In the Mitra 15 minicomputer, the most important chips were again Texas. For good measure, we also looked inside a Micral (the first microcomputer, built in 1972, based on the Intel 8008) made by an independent small company, R2E : without exception, all its logic components were American, and of 200 circuits 119 were from Texas The British "national champion" acted in just the same was as CII: in order to remain competitive against the Americans, ICL used Texas Instruments circuits rather than those of its compatriot, Ferranti..

In a report dated 1974 the Dйlйgation а l'informatique said: "At the same time as its notorious deficiences in production and even development of Schottky TTL logic circuits, Sescosem has had good results in the field of bipolar memories and now has a good, though not perfect, command of several MOS technologies. The general outcome is, however favourable, since the fraction of components used by CII that are of French origin, almost zero at the start of the Plan Calcul, is over 50% [at the end of the second Plan Calcul] (Our italics) The author of this report does not say whether the percentages quoted refer to all components or only to integrated circuits; it seems to mean the latter. The text does not give any percentages in terms of value. Quoted by P. Audoin (1988: 36). A much less optimistic estimation, in a 1975 pamphlet published by CII unions, stressed that Sescosem supplied only 5% of CII total consumption of s.c.". This certainly shows progress, but seven years had gone by; given the cumulative nature of innovation in the semiconductor industry, for France to recover from its shortcomings had become a practical impossibility.

As emphasised by John Zysman, “Although learning-curve economies occur in alll branches of industrial production, they are critical to an understanding of the component business beacause they have a peculiar impact on business strategy in the active component industry. The cost of producing semi-conductors appears to decline by 20-30% every time the total number of units produced doubles” (Zysman 1982). The refusal of the government to associate Bull with the Plan Calcul in 1966 made CII a company of sub-critical size (one-third the size of Bull), whether in terms of output or of negotiating power with the component producers (Mounier-Kuhn 1993). The hoped-for driving effect of informatics on the French semiconductor industry never materialised to any decisive extent.

Conclusion

At the start of the 1960s there were more than ten companies making computers in France. These were generally in very small runs, a few tens of units and in some cases only a few units -- apart from IBM and Bull. Nevertheless, this group of firms constituted a respectable market for the component producers, of the order of a million transistors annually, and increasing rapidly.

The major French electronics group, CSF, seems to have had no very coherent policy in this field. After refusing for a long time to invest in digital computing, it created CAE, with licences to build American machines: this was entirely contrary to its traditions of relying on its own R&D to create innovative products. With COSEM encountering difficulties that delayed the necessary moves first to silicon transistors and then to integrated circuits, CAE turned to Texas Instruments; consequently there was little synergy between computer design and component manufacture within the CSF group. This is a point of major importance, which should be looked into carefully in any future comparative study of the international scene: beyond the case of CSF, why did the Japanese electronic giants -- NEC, Toshiba, Fujitsu -- succeed in informatics, whereas the corresponding American and European (especially French) companies abandoned this, one after the other The one exception was Siemens, who however have experienced many setbacks. The conclusions we have come to as a result of studying the French companies confirm what was said by Franco Malerba (1985), with a few qualifications concerning:

- the role of the State, which was more positive than one might have thought in the nascent French computer market; the public sector bought about 60 of the 120 machines installed between 1955 and 1959

- the role of IBM, whose French subsidiary bought its components mainly from European producers up to the end of the 1960s.

- the absence of "new entrants", enterprises going into semiconductor production and renewing the industrial fabric. Malerba reckons that the absence of companies like Motorola and Intel largely accounts for the decline of the European electronics industry. This is a persuasive argument, but seems to us over-simplified. There was indeed a French "new entrant", SILEC, formed in 1956 by a Thomson engineer, to make silicon components; this was a success story and is now a piece of forgotten history.Our impression is not that there was a lack of entrepreneur spirit in France, but that the State always tended to support existing companies and sectors, rather than the nascent ones.?

Was the failure of Bull, CGE and CSF to reach agreement in 1963 an historic opportunity missed? The three companies had sketched out an alliance, in an attempt to overcome Bull's crisis; but Bull, who found no concrete help from its French partners, preferred to join General Electric in 1964 -- motivated in part by the hope of finding a competitive supplier of components. Would not a strong alliance of the three French companies, supported by the State, have created an above-critical mass which, whilst certainly less than IBM, would nevertheless have been large enough to persuade the French semiconductor industry, and especially COSEM, to turn its attention to the components with the greatest potential? This is wishful thinking, since neither CGE nor CSF were willing to attack the commercial computer market.

Without this agreement, French informatics had no consolidated front to present against the American component suppliers; the latter steadily increased their domination both by selling directly to the computer manufacturers and by licensing their processes to the French producers, who then became "second source" suppliers. The Plan Calcul and the Plan Composants had little effect; they were too late and lacked any serious strategy -- they aspired to produce business computers, while excluding Bull. It could even be claimed that the two plans were mutually inconsistent, since the national champion (CII) they set up, starting with a heavy commercial handicap in front of IBM and Digital Equipment, had either to use American integrated circuits in its products or be completely uncompetitive The same behaviour, resulting from the same causes, was observed at the same time at ICL and Olivetti..

Does this mean that planning is basically wrong, and that we should trust only the unregulated market to manage our industry and promote innovation ? No. The Plan Calcul failed mainly because its initial managers devoted more attention to politics and not enough to real industrial and technical strategy. In fact, the Plan Calcul was neither seriously planned nor calculated, if we consider the organization of research, production and sales. The government should not have picked up “national champions” among big, established companies with political connections - but among the smaller companies which were successfully developing and selling innovative minicomputers. It should have helped the innovators, the winners, instead of pouring the taxpayers' money into mismanaged firms. It should have maintained competition between them and let the market decide who was to be the “national champion” - instead of decreeting this from a Ministry office. In other branches of computing activities, the governmental policy was more subtle, more modest, more appropriate: The French software industry has prospered in a context of free competition with contracts from State administrations; the French telecommunication industry, which was almost unexistent 40 years ago, has become a pionneer in digital telecom and networks, based on long-term research done at the Centre National des Telecommunications; the smart card, patented in 1974 by a young inventor, soon supported by France Telecom, has become an industry and contributed to change our lives.

Literature

Allиgre, M. (1969) Interview in Йlectronique Actualitйs, 17 janvier.

Audoin, P. (1988) “Le Plan Calcul Franзais (1966-1974)” Actes du colloque sur l'histoire de l'informatique en France, INPG, Grenoble.

Bashe, C. J., Johnson L. R. & Palmer J. H. (1991) IBM's 360 and Early 370 Systems, Cambridge, Ma., MIT Press series in the History of Computing.

Borne, J. & Audebert M. (1961) “Circuits de calcul utilisant des transistors et des tores magnйtiques”, Acta Electronica, vol. V, n° 32.

Botelho, A. J. (1990) “The State and the political construction of informatics in France and Brasil” Actes du 2e Colloque sur l'Histoire de l'Informatique en France, CNAM, Paris.

Cristini, G. & de Lamaziиre A. (1970) “Le parc franзais”, 01 Informatique, septembre.

Danzin, A. (1958) “Le ferrite, matйriau clй de l'йlectronique” Tйlonde n° 3 p. 5.

Delamarter (1986) Big Blue - IBM's Use and Abuse of Power, MacMillan, Londres.

Dreyfus-Alain, B. & Thorn Y. (fйvrier-mars 1964) Voyage d'information aux Йtats-Unis - Rapport de synthиse Document interne SEA.

Jacq, F. (1991) Un regard sur la science et l'industrie dans la France d'aprиs-guerre -- Le Laboratoire d'йlectronique et de technologie de l'informatique : une aventure scientifique grenobloise (1955-1980) Centre de sociologie de l'innovation de l'Ecole des Mines, Paris.

Kilby, J. (1966) (Texas Instruments) Confйrence au Salon des Composants de Paris Йlectronique Actualitйs, 15 mars 1966.

Kranakis, E. (1990) “Early Computers in the Netherlands”, Actes du 2e colloque sur l'histoire de l'informatique en France, CNAM, Paris.

Leclerc, B. (1990a) “Les premiиres machines йlectroniques Bull : du Gamma 3 au Gamma E.T.”, Actes du 2e colloque sur l'histoire de l'informatique en France, CNAM, Paris.

Leclerc, B. (1990b) “From Gamma2 to Gamma ET: The Birth of Electronic Computing at Bull”, Annals of the History of Computing vol.12/1.

Malerba, F. (1985) The Semi-Conductor Business, University of Wisconsin Press.

Mounier-Kuhn, P.-E. (1989) “Bull - A Worldwide Company Born in Europe” Annals of the History of Computing vol. 11/4.

Mounier-Kuhn, P.-E. (1989), “The Institut Blaise-Pascal (1946-1969) - From Couffignal's Machine to Artificial Intelligence” Annals of the History of Computing vol. 11/4.

Mounier-Kuhn, P.-E. (1990) “Bull : 70 ans de traitement de l'information” Actes du 2e Colloque sur l'Histoire de l'Informatique en France, CNAM, Paris.

Mounier-Kuhn, P.-E. (1993a) “Les collections d'informatique en France” Revue du Musйe des Arts et Mйtiers Paris, n° 2.

Mounier-Kuhn, P.-E. (1993b) “Le Plan Calcul, Bull et l'industrie des composants : les contradictions d'une stratйgie”, Actes du 3e colloque Histoire de l'informatique, INRIA, Sophia-Antipolis. (re-published in Revue historique, 1995, vol. CCXC n° 1).

Mounier-Kuhn, P.-E. (1994) “Product Policies in Two French Computer Firms: SEA and Bull (1948-1964)” Information Acumen - The Understanding and Use of Knowledge in Modern Business, L. Bud-Frierman, йd., Routledge, London 1994.

Mounier-Kuhn P. & M. Grossetti (1995) “Les dйbuts de l'informatique dans les universitйs - Un moment de la diffйrentiation gйographique des pфles scientifiques franзais” Revue Franзaise de Sociologie, 1995-2.

Mounier-Kuhn, P.-E. (1995) “Un exportateur dynamique mais vulnйrable : La Compagnie des Machines Bull (1948-1964)”, Histoire, Economie et Sociйtй, 1995 n° 3.

Pestre, D. (1984) Physique et physiciens en France 1918-1940, Editions des Archives contemporaines, Paris.

Pestre, D. (1989) Intervention au Colloque sur l'Histoire du CNRS, Cahiers pour l'Histoire du CNRS, n° 6 p. 43-44.

Rajchmann, J. A. (1952) “Les systиmes matriciels de mйmoire”, (3 juillet 1952) est publiйe dans L'Onde йlectrique n° 32

Raymond, F.-H. (1965) Ensemble de calcul universel 1968, rapport de synthиse secret au Ministиre des Armйes (DRME), rйf. RT 260.006/8, 28 juillet 1965, 16 p.

Raymond, F.-H. (1967) Dossier sur les apports incorporels а CII, Observations sur le dossier CAE-Analac, private archives.

Raymond, F.-H. (1989) “SEA: An Adventure with a Sad Ending”, Annals of the History of Computing vol.11/4.

Vernay, J. (1988) Chroniques de la Compagnie IBM France IBM йd.

Zysman, J. (1982) Political Strategies for Industrial Order State, Market and Industry in France, University of California Press.

Размещено на Allbest.ru