What is the significance of high-speed computation for the sciences? How far does it result in a practice of simulation which affects the sciences on a very basic level? To offer more historical context to these recurring questions, this paper revisits the roots of computer simulation in the development of the ENIAC computer and the Monte Carlo method.With the aim of identifying more clearly what really changed (or not) in the history of science in the 1940s and 1950s due to (...) the computer, I will emphasize the continuities with older practices and develop a two-fold argument. Firstly, one can find a diversity of practices around ENIAC which tends to be ignored if one focuses only on the ENIAC itself as the originator of Monte Carlo simulation. Following from this, I claim, secondly, that there was no simulation around ENIAC. Not only is the term ‘simulation’ not used within that context, but the analysis also shows how ‘simulation’ is an effect of three interrelated sets of different practices around the machine: (1) the mathematics which the ENIAC users employed and developed, (2) the programs, (3) the physicality of the machine. I conclude that, in the context discussed, the most important shifts in practice are about rethinking existing computational methods. This was done in view of adapting them to the high-speed and programmability of the new machine. Simulation then is but one facet of this process of adaptation, singled out by posterity to be viewed as its principal aspect. (shrink)

In response to the so-called ‘software crisis’ of the late 1960s, many approaches were proposed to turn software engineering and programming into more systematic disciplines, to turn an art into a science. This paper studies one popular example often used in these proposals, the computation of a list of primes, to discuss some salient features of the proposed programming paradigms. It also looks at the actual implementation in the early 1970s of the prime program on a time-sharing system and on (...) a complex scientific computer. Confronting theory with practice uncovers what the programming paradigms fail to grasp: the interaction with the user and the interaction with the machine. (shrink)