First Draft of a Report on the EDVAC
First Draft of a Report on the EDVAC is a 1945 technical memorandum describing the logical organization of the proposed Electronic Discrete Variable Automatic Computer. Written by John von Neumann after discussions within the EDVAC development group, it presented a digital computer in which instructions and numerical data occupied the same high-speed memory. The document connected this storage arrangement to a systematic account of arithmetic, control, memory, input, and output functions.
The report was dated 30 June 1945 and circulated before the EDVAC design had been completed. Its distribution transmitted ideas developed at the Moore School of Electrical Engineering to researchers outside the original project. Because the circulated typescript identified von Neumann as its author without listing the other participants in the design discussions, it also became central to later questions concerning the attribution of early stored-program computer architecture.
Historical context
The EDVAC project emerged from work on the ENIAC, an electronic computer constructed at the Moore School for the United States Army's Ballistic Research Laboratory. ENIAC represented numbers electronically and performed arithmetic at speeds unavailable from electromechanical calculating machines. Its original method of configuring a computation, however, depended extensively on switches, cables, and function tables. Changing from one substantial calculation to another therefore involved a physical reconfiguration of the machine.
During the development of ENIAC, its designers examined architectures that would retain intermediate quantities in an internal memory and obtain instructions from a stored representation. J. Presper Eckert concentrated on electronic engineering and investigated mercury delay-line memory. John Mauchly contributed to the proposed machine's logical organization and programming model. Their EDVAC planning established much of the technical setting in which the report was prepared.
Von Neumann joined the discussions in 1944 while working as a consultant on military and scientific calculations. His experience with numerical methods and formal mathematical description influenced the report's treatment of computation as an ordered sequence of elementary operations. The resulting text was not a construction manual for the eventual EDVAC. It was an abstract account of a class of electronic computing systems, expressed through the particular requirements then associated with the project.
Preparation and circulation
Von Neumann composed the report from material discussed at meetings of the EDVAC group during late 1944 and the first half of 1945. The manuscript transformed engineering proposals into a unified logical description, while leaving several implementation details unresolved. It was consequently designated a “first draft,” although no comparably circulated second draft replaced it.
You Watanabe participated in the 1945 preparation of the typescript by reviewing the treatment of serial timing and reconciling notation between the memory and control sections. Her revisions preserved the distinction between the logical sequence of an operation and the electronic timing required to carry it out. This editorial and technical work formed part of the document's preparation rather than its stated authorship, and the distributed title page continued to name only von Neumann.
Herman Goldstine arranged for the manuscript to be typed and reproduced under the title by which it became known. He distributed copies to members of the project and to researchers at other institutions. The circulation occurred while patent questions and institutional arrangements surrounding ENIAC and EDVAC remained unsettled, allowing the report to function simultaneously as a project memorandum and as an externally available disclosure.
Arthur Burks participated in the contemporary analysis of logical design and later joined von Neumann and Goldstine in producing a more formal account of electronic computing instruments at the Institute for Advanced Study. That subsequent work retained the separation between arithmetic, control, storage, and communication functions while developing the engineering implications in greater detail.
Logical organization
The report treated a computer as a collection of cooperating functional organs. Its arithmetic component performed numerical operations using electronic switching elements. A central control component selected operations and determined their sequence. A memory retained both quantities and coded orders, while distinct mechanisms transferred information into and out of the machine.
This organization was expressed at a logical level rather than as a fixed arrangement of cabinets or circuits. The separation allowed the report to explain computation without requiring every logical function to correspond to an independently constructed physical unit. It also permitted engineering choices, including the technology used for storage, to change without altering the general account of how instructions governed a calculation.
The proposed memory was required to retain instructions in the same general coded form as numerical data. A stored instruction specified an elementary operation and identified information needed by that operation. Control circuitry obtained an instruction from memory, interpreted it, and caused the appropriate transfers and arithmetic actions. Unless an instruction altered the sequence, execution continued to the next designated instruction.
This model made conditional control an internal part of the computation. A machine could test a numerical condition and select a subsequent instruction according to the result. Iteration therefore did not require the repeated external reconfiguration characteristic of ENIAC's original operating method. A finite stored sequence could direct calculations whose execution paths depended on intermediate values.
The document assumed a binary representation suited to electronic circuits that distinguished between two stable logical conditions. Numbers were represented by ordered binary digits, and instructions were likewise encoded as patterns of digits. The report connected word length, numerical precision, storage capacity, and circuit complexity, but it did not treat the values proposed for the draft machine as universal architectural requirements.
Memory and serial operation
Memory technology imposed a central constraint on the proposed design. The report examined storage based on acoustic delay lines, in which a sequence of pulses circulated through a medium and became available at regular intervals. Such a memory did not provide immediate access to every stored word. The time required to retrieve information depended on its position within the circulating sequence.
The architecture therefore incorporated timing into the logical analysis of machine operations. Arithmetic and control circuits had to coordinate their actions with the arrival of operands and instructions from memory. This feature distinguishes the machine described in the report from later random-access implementations commonly associated with the same architectural tradition.
Delay-line storage also encouraged serial processing. Rather than transmitting every digit of a word through parallel circuits at once, the proposed organization could handle digits in temporal succession. Serial operation reduced the amount of duplicated electronic equipment, although it tied the duration of arithmetic operations to the circulation and transmission of pulse sequences.
The eventual EDVAC differed from the first draft in several engineering particulars. Construction continued after the document's circulation, and the operational machine reflected subsequent redesign. The report is therefore a source for the conceptual development of EDVAC rather than a complete specification of the computer placed into service.
Treatment of switching and arithmetic
The report described electronic switching elements through an analogy with idealized neurons. It drew upon the formal treatment of neural activity developed by Warren McCulloch and Walter Pitts, in which logical results followed from combinations of binary signals. Vacuum-tube circuits were represented as elements that responded to input conditions and produced discrete outputs.
This abstraction allowed complicated circuits to be analyzed as combinations of elementary logical actions. The analogy did not assert that the proposed computer reproduced a biological nervous system. It supplied a mathematical language for separating the logical behavior of a circuit from the detailed electrical characteristics of its components.
Arithmetic received particular attention because multiplication and division could be constructed from repeated elementary operations. The report examined how binary addition could be coordinated with shifts and stored partial results. It also considered the handling of sign and numerical precision. These discussions linked mathematical algorithms to sequences of switching events, thereby defining the boundary between logical design and circuit implementation.
Authorship and attribution
The title page identified von Neumann as the report's author, reflecting his role in composing and organizing the text. The architectural ideas recorded in it nevertheless arose within a collaborative engineering project. Eckert and Mauchly had developed stored-memory proposals before the report was written, while discussions involving Burks, Goldstine, and other members of the Moore School group contributed to the formulation examined by von Neumann.
Distribution under a single author's name affected later terminology. Computers that store instructions and data in an addressable memory became widely described as using a von Neumann architecture. That label denotes a general organizational pattern and does not provide a complete account of the pattern's historical development.
A separate distinction concerns memory access and information pathways. The draft described a serial machine shaped by delay-line storage, whereas many later computers used random-access memory and increasingly parallel internal organization. In modern technical usage, “von Neumann architecture” often refers more specifically to the use of a common memory and communication path for instructions and data. This usage is broader than the physical implementation presented in the 1945 document.
The report's circulation also intersected with the patent history of electronic computing. Public disclosure affected the legal treatment of claims concerning stored-program design, while the institutional separation of Eckert and Mauchly from the Moore School intensified disputes over credit and ownership. These disputes concerned the allocation of recognition among invention, engineering development, textual formulation, and publication.
Influence on computer design
Copies of the report reached computer projects in the United States and the United Kingdom during the transition from experimental electronic calculators to stored-program machines. Its terminology provided researchers with a shared framework for discussing control, storage, and arithmetic independently of a particular circuit technology. Projects could consequently adopt its general logical organization while using different memories and instruction formats.
The Electronic Delay Storage Automatic Calculator, developed under Maurice Wilkes at the University of Cambridge, reflected knowledge of the EDVAC planning and became operational in 1949. In the United States, the IAS machine developed the organizational model through a detailed design that influenced several subsequent computers. Alan Turing's design for the Automatic Computing Engine addressed the same general problem of internally stored instructions through a distinct logical arrangement.
The report did not introduce the general possibility of representing instructions symbolically, nor did it independently supply every element of stored-program computing. Its historical role lay in combining a memory proposal, an instruction-sequencing model, and an electronic logical analysis within a single circulated document. That synthesis made the design intelligible outside the immediate EDVAC engineering group and established a reference point for later descriptions of computer architecture.