ENIAC and Its Contemporaries Meet the “Stored Program Concept”

Having explored ENIAC’s actual use and the programs it ran the authors shift to a more abstract analytical level. Previous discussion of the invention of the modern computer has focused on the “stored program concept” as the crucial innovation setting modern computers apart from their more limited predecessors. The authors explore the origins of this phrase and its changing meaning over time. They look in detail at a 1944 document produced by J. Presper Eckert and sometimes claimed as a first statement of this concept, showing that it actually describes an electronic desk calculator. The authors summarize ENIAC’s capabilities after conversion and to compare these on both practical and theoretical levels with the 1945 EDVAC design and with several other early computers. This supports a balanced appraisal of the senses in which the converted ENIAC did and did not constitute an initial implementation of the key ideas from the 1945 design. The chapter argues for an appraisal of early computers better grounded in the historical realities of documented use, and against a widespread fixation on the notion of “universality” based on a school of theoretical computer science that gained prominence years later.

Converting ENIAC

In spring 1947 a project was launched to convert ENIAC to run code written in the new from introduced with the 1945 “First Draft of a Report on the EDVAC.” This was intertwined with the planning of Monte Carlo calculations for Los Alamos. Adele Goldstine worked with a team of contractors led by Jean Bartik and a group of Aberdeen employees under Richard Clippinger to develop a succession of planned “set-ups” to implement a new control mechanism and vocabulary of general purpose instructions for ENIAC. Our analysis focuses particularly on the relationship of this work on concurrent efforts by von Neumann’s team on the design of the Institute for Advanced Studies computer and a series of related reports on programming methods. Accounts by participants and historians have differed dramatically in assigning credit for the conversion and on such basic facts as when the conversion was implemented and what version of the design was used. The conversion was finally implement in March 1948 by Nick Metropolis (of Los Alamos and the University of Chicago) using a variant design he formulated with Klara von Neumann. At this point ENIAC became the first computer ever to execute a program written in the “modern code paradigm.”

EDVAC and the First Draft

This chapter charts the rapid evolution of thinking about programming and computer architecture among members of the ENIAC team from 1944 onward, as what is usually called the “stored program concept” was formulated with John von Neumann and presented in the “First Draft of a Report on the EDVAC.” Use of archival sources makes this more specific and rigorous in documenting this process than any previous published account, presenting the new approach as an evolution of, and response to, the original ENIAC programming method. The ideas present in the “First Draft” are clearly explained and separated into three distinct clusters: the “EDVAC hardware paradigm,” the “von Neumann architecture paradigm,” and the “modern code paradigm.” The chapter finishes with an exploration of initial understanding and reception of these ideas, reconstructing the late-1940s consensus on what was important about the new approach and why.

ENIAC Goes to Monte Carlo

Traces the series of Monte Carlo simulations run on ENIAC from their genesis in January 1947 exchanges between John von Neumann, Robert Richtmyer, and Stanislaw Ulam through the completion of detailed planning work for the initial batch of calculations in December 1947. Close attention to successive drafts illuminates the process by which John and Klara von Neumann worked with Adele Goldstine to transform the former’s outline plan of computation into a fully developed flow diagram documenting the flow of control and manipulation of data for a program written in the new style.

ENIAC Arrives at the Ballistic Research Lab

In 1947 ENIAC was unpacked and installed at the Ballistic Research Laboratory in Aberdeen, Maryland. This chapter explores its first fifteen months there, beginning with some context about the mission and organization of the laboratory. It depicts ENIAC as one of several automatic computers in use there, seen at the time as a device with particular strengths and weaknesses but broadly comparably to electromechanical relay computers in its abilities. Drawing on a previously neglected source, the ENIAC Operations log, it reconstructs the challenges and frustrations involved in attempting to work with electronic computers during this era. Discussion focuses on the two main known problems it ran during this period – interpreting flight data from V2 missiles and creating tables for the analysis of statistical outliers.

Imagining ENIAC

Explores the origins of ENIAC, beginning with its two primary inventors John W. Mauchly and J. Presper Eckert. Examines ENIAC’s original 1942 and 1943 proposals, the process by which the machine was approved, and the initial assembly of a team to work on the project. Documents connections between ENIAC and earlier work on differential analyzers, including a machine constructed during the 1930s and a 1940 proposal by Irven Travis for the creation of a digital electronic version. Analysis focuses particularly the relationship between the problem for which ENIAC was commissioned, the computation of artillery trajectories, and the proposed design. It concludes with a description of early planning work to determine the basic capabilities of ENIAC’s accumulators, its standard building block.

ENIAC Tries Its Luck

As soon as Metropolis had completed the initial configuration of ENIAC for the new programming method, and before it was working properly, Klara von Neumann arrived to help. She had taken the leading role in converting the flow diagrams into program code, and together they worked around the clock for several weeks to get both program and machine into a usable state and to shuffle tens of thousands of cards in and out of it during Monte Carlo simulation of each exploding fission bomb. This chapter integrates the narrative of this initial “run,” of and a second batch of calculations carried out in late-1948 with analysis of the structure of the program itself. It finishes with an exploration of further Monte Carlo work run on ENIAC, including reactor simulations, simulation of uranium-hydride bombs, and in 1950 simulation of the “Super” concept for a hydrogen weapon.

Structuring ENIAC

This chapter explores the development of ENIAC’s overall architecture and control method. This was shaped, to a degree that has not previously been recognized, by an early and very detailed exploration led by Arthur W. Burks of how the machine could be “set up” to calculate shell trajectories, the task for which it was commissioned. Programming ENIAC was not, as has often been asserted, an “afterthought” to its design and construction. Discussion is focused in particular on the development of its master programmer unit, used to control sets of nested loops. Although it is widely believed that the ability to change the course of a computation based on results so far obtained (later be conceptualized as a conditional branch) was added to ENIAC late in its development, we show that this capability was planned for early on and that its eventual implementation as a capability of the master programmer reflected a distinct approach to the structuring of automatic computation shaped by the team’s work on the trajectory computation problem.

Introduction

Introduces ENIAC and sketches its accepted place in the history of computing as a candidate for the disputed honor of “the first computer,” or as the “first general purpose electronic digital computer.” The authors argue that both views simplify ENIAC’s complexities by reducing it to a single point on a historical trajectory. Instead they introduce a number of other perspectives developed in the book: ENIAC as a material artefact, ENIAC at the origin point of computer programming, ENIAC as a site for technical analysis, and ENIAC as an object of contested historical memory.

Conclusion

This concluding chapter engages with key ideas from treatments of ENIAC by authors such as Jennifer S. Light and Nathan Ensmenger to explore ENIAC’s legacy in areas such as computer programming, computer center organization, and computer simulation. It is inspired by the treatment of ENIAC’s historiographic role by the late Michael S. Mahoney in his provocative essay “The Histories of Computing(s).” Mahoney challenged the assumed centrality of ENIAC to all history of computing narratives. Having shown in the previous chapter how this perceived position developed over time we can now make a balanced reappraisal, arguing that ENIAC’s influence was profound and underappreciated in some areas but has been overstated or misinterpreted in others. One important topic is the idea that ENIAC’s operators were the “first computer programmers,” a description that distorts their actual relationship to the machines.