Birth of a science

Informatics is a child of this century. Birth of a science - doesn't that sound pompous? Can science really be born at all? Despite all the scepticism, the saying that computer science is a baby of the 20th century is quite appropriate. In 1955, not even the name was coined. What led to its rapid rise and sustained survival? No doubt, the development of electronic circuits for information processing, a byproduct of World War Two, found its continuation in further civil and military applications. Smaller and robuster components were needed. The invention of the transistor and microminiaturization of transistor circuits led to a dramatic improvement in the price-performance ratio of computers. Microelectronics proved to be a major impulse for informatics.

The rudiments

To continue the metaphor, informatics - or better, what we mean today under this word - was conceived at the very beginning of our history of science. 2300 years ago, Euclid taught a method for the determination of the greatest common divisor of two numbers, today called Euclid's algorithm. The word algorithm points back to Al Chwarizmi, who lived at the court of the Caliphs of Baghdad and, 1200 years ago, taught mechanical rules for calculations with Indian ("arabic") figures and with numbers composed from them. This started a comprehensive variety of calculations, mainly in the field of astronomy. Calculations were performed by hand according to stringent drill rules; in the 17th century, primitive adding machines (Wilhelm Schickard, Blaise Pascal) came into existence. Subsequently, W. Leibniz mechanized even multiplication - Schickard was stuck halfway in doing this. Although initially mechanical difficulties occurred, this led in the 19th century to successfully operating, hand-driven four-function machines, made by craftsmen (Thomas), and around the turn of the century to industrial fabrication (Burkhardt, Odhner, Steiger). At about the same time, Hollerith and Powers invented machines for the evaluation of statistical records, Burroughs and Fisher rationalized offices with the help of accounting machines, and Ritty and Patterson rationalized sales departments with the help of cash registers. In this way, the general public was around 1930 confronted with isolated events of mechanization, on which the idea of the common computer, one solid foundation of informatics, was to be based. But other grass roots of informatics go further back. The importance of number representation for arithmetic was realized early. For a long time, the sexagesimal system and the decimal system stood side by side. Leibniz brought a new aspect into the development. Dissociating himself from the then prevailing decimal system, he propagated the dual system. At first, this met with little or no success. However, in the 20th century, the Leibniz calculation method in the dual system proved ideal for electronic technology.

Leibniz also liberated the term "algorithm" from its limitation to figures and numbers and expanded it to a game with objects of any given meaning, a game based on specified, fixed rules. The notion of "algorithm" in this general sense defines precisely what we have come to know as data processing or information processing. It is another solid foundation of informatics which, in this sense, goes back to Leibniz. Cryptology, a field that had also attracted Leibniz, was based on a very general concept of algorithm, too. A further field was the mechanizing of simple syllogisms and propositional functions, which was attempted by Marquand and Peirce, around 1890 Leibniz had already experimented with the techniques of encoding of concepts. All this was premature.

Control was something Leibniz did not yet incorporate. Individual inventors of clocks and mechanical androids sparked an idea which was further explored by Charles Babbage and the Spaniard Leonardo Torres y Quevedo to develop total automation of the calculation process. The breakthrough was approached with the invention of fully automatic punch-card machines and book-keeping machines. Moreover, mechanical desk calculators were developed further, becoming electrically powered and performing the arithmetic operations fully automatically. The crowning achievement was the introduction of an automatic square-root function in the Friden calculator.

The thirties

Inevitably, at this time questions of computer technology were in the forefront: realization of the necessary basic operations with the means available and optimization of the computation process. Quite a number of solutions worked out in those days of mechanical and relay computers are today of historical interest only. Nevertheless, something was lasting: Almost unnoticed, concepts and principles for modelling information processing emerged. Computer science existed in embryo when in the '30s restless times began. Soon unforeseen developments emerged. Mathematical logic, plagued by paradoxes, refined the concepts "provable" and "computable", leading to the fundamental theorems by Kurt Gsdel on the limits of formal provability and to the Gedankenmaschine of Alan Turing, which was intended to be the prototype of a "universal machine". And although extensive relay circuits had been technically feasible for fifty years, only around 1935 were machines built - by Konrad Zuse

in Germany, by Wallace J. Eckert, by George R. Stibitz, and by Howard Aiken in the U.S.A. - that could do more than the mere mechanical calculators that carried out single arithmetic operations after pushing buttons. Full automation finally led to machines that performed whole calculations without assistance and were usable for a multitude of different algorithms - freely programmable program-controlled machines. Moreover, in some places it took a long time to break with the decimal system and with the mechanical cogwheels seemingly inseparably connected with it; the decimal system was imitated by bulky decimal ring counters. Only Konrad Zuse in Germany went the straight way; he freed himself from the decimal system and found an interesting technology for mechanical calculation without cogwheels before passing over to electromechanical relays.

The breakthrough

Not more than ten years later, the breakthrough was accomplished by the first electronic versions of such calculating machines - Helmut Schreyer in Germany, John P. Eckert and John Mauchly in the U.S.A., and Thomas Flowers in Great Britain were the pioneers - not to forget the outsider John Vincent Atanasoff, who skipped the relay phase. To continue the metaphor, the labour pains started. Electronics allowed an immense increase in speed (and later, after semiconductor technology was introduced, unforeseen reliability). The birth was accompanied by a complete liberation from the constraints of doing calculation with numbers: John von Neumann and his group demonstrated in 1946 with the stored program machine EDVAC how in principle algorithms with arbitrary (binary coded) objects were programmable; logicians soon demonstrated the universality (in the sense of Turing) of such machines. which came to be called computers. The only restriction on the objects the algorithm struggled with - (a countable number of) numbers, truth values, or any other, even composite objects - was that they could be somehow coded. The universality of binary coding became also the basis of the Plankalkül of Konrad Zuse, thus helping an old idea of Leibniz into full bloom. Quite soon, first applications followed, exploiting fully the developed power of algorithms, for example programs for mechanical execution of differentiation of a formula (Kahrimanian and Nolan 1953) or heuristic proof programs (Gelernter and Rochester 1958); in 1959, with the programming language LISP, John McCarthy could advocate the general use of recursive programs. In this stormy time, when new ground was broken, preparatory work, e.g., in mathematical logic, was sometimes not utilized for lack of time. On occasions, the education the pioneers had been given closed doors. Whatever harm this did, it also helped the budding science of informatics to become self-reliant. It was formed as a scientific discipline by agglomerating many single subjects around a core concern. This centerpiece of informatics was research into the possibilities of describing algorithms with the help of programming languages, including their complexity, their efficiency with respect to given algorithmic machines and finally realization by new machine architectures corresponding to prevailing technological progress. This is so even today; we shall come back to this in sect. 3.

Friedrich L. Bauer
Wilfried Brauer
Eike Jessen
Manfred Broy