Tuesday, November 9, 2010

Pierre Mounier-Kuhn, "L'informatique en France", PUPS, 2010

Translated from a French version

This is the first volume of a monumental work that Pierre Mounier-Kuhn dedicates to the history of computing in France. It addresses research and University. A second volume will deal with the behavior of manufacturers, a third with the role of the state.

It will probably be necessary for Mounier-Kuhn to devote a fourth volume to the computerization of institutions, to what happened on the side of usages, their requirements and their relationships with scientists, suppliers and the state.

Like any historical research, this one was constrained by the availability of archives and hence it considers what happened before 1975. But it is at this time that the computerization of institutions and society begun: the notion of information system emerged in the early 70s, when the requirements of the semantics of data passed before those of the calculation which they condition under the principle garbage in, garbage out.

Mounier-Kuhn writes on the computerization of France, but he shows that our country, delayed by the circumstances of the German occupation, has not been a pioneer in this field. Computer technology had its source in the U.S. (and Britain), and the ideas that we French have had on compilers, operating systems, databases, networks etc. were only second hand, notwithstanding some individual exceptions.



It is therefore difficult to describe the computerization of France without going back to the reservoir of ideas and American innovation from which flowed to us a relatively thin thread. How indeed to evoke programming languages without describing their birth and that of compiling with Fortran in 1954 (cf. L'émergence des langages de programmation) and the problems their design faced? The same question arises about operating systems, databases, transaction monitors, etc. But then that's another story to write, upstream of the history of computing in France which, by comparison, may seem minuscule: this difficulty is inherent in the project of Mounier-Kuhn.

His book demonstrates the ambiguity of the relationship between computing and mathematics. In the beginning computing is a tool for calculation, and is therefore connected to applied mathematics. Computer science tries to win legitimacy in scientific research and universities, but it meets with the contempt for applications from the "Bourbakists" which, obsessed with "purity", have transformed mathematics in an instrument of mental torture.

True, algorithmic poses formidable mathematical problems: those that floating point computation arises seem even insurmountable [1]. But here as in other disciplines the effort, unsuccessful indeed, to conciliate with "pure" mathematicians (or at least to give to computer science the prestige of "pure" Mathematics) has pushed the formalization well beyond what is required by algorithmic.

A key difference separates computing from mathematics: while they explore a logical world which is indifferent to the passage of time as to its emergencies, computing is action-oriented [2].

There is something pathetic in the attempt to mathematize computer science: it is painful to see this discipline, which being oriented towards action is scientific in the strict sense of the term, imitate laboriously mathematics in an attempt to acquire scientific reputation. It's away from his own purpose, which is essentially practical, and it does indeed reinforce the hostility felt by "pure" mathematicians.

Behind the conflict between mathematics and computing lays the more general question of the criteria by which one can evaluate the scientific nature of a discipline. Those who believe that science depends on rigorous demonstrations forget that each science is suspended to principles, or axioms, which are not demonstrated and which define its realm. The realism of mathematics is that of thought subject to the sole rule of non-contradiction. The realism of computing is that of action, a meeting of nature with intentions oriented by values.

Mounier-Kuhn describes in detail the painful resulting institutional vicissitudes. To the exceptional lucidity of a François-Henri Raymond contrast the careerism and pedantry of many others, as well as dreams and fantasies coming from a tote-bag of concepts: cybernetics, artificial intelligence, systems theory, fuzzy logic, complex thinking, even information theory.

This history is already old and computerization has transformed the landscape. But the institutions established before 1975 are still there and their priorities remain the same: many students never hear of computer information systems (this term does not appear in the training of Ecole Polytechnique) or of the semantic engineering that builds their foundation. They have to swallow equation on equation, and it does not prepare them to understand computing and even less computerization.

____________________ ____________________

[1] « Many serious mathematicians have attempted to analyze a sequence of floating point operations rigorously, but have found the task so formidable that they have tried to be content with plausibility arguments instead » (Donald Knuth, The Art of Computer programming , Addison-Wesley, 1998, vol. 2, p. 229).

[2] « In mathematics we are usually concerned with declarative (what is) descriptions, whereas in computer science we are usually concerned with imperative (how to) descriptions » (Harold Abelson and Gerald Jay Sussman, Structure and interpretation of computer programs , MIT Press 2001, p. 22).

No comments:

Post a Comment