Diff — EDVAC report

The **First Draft of a Report on the EDVAC** — universally called the **EDVAC report** — is a 101-page typescript written by **[[John von Neumann]]** and distributed on **30 June 1945**. In fewer than a hundred pages it described the architecture of a computer that would keep its *instructions in the same memory as its data* — the design almost every computer built since has followed. It is arguably the most influential document in the history of computing, and also one of the most contested.

# What the report actually says

The report describes **EDVAC** — the *Electronic Discrete Variable Automatic Computer*, the Moore School's planned successor to ENIAC. Its machine had:

- A **memory** of mercury delay lines — 32-bit words, with a design goal of 8,192 — holding numbers **and instructions together**: the stored program

- A central **arithmetical organ** — the CA, today's ALU/CPU — that fetched instructions from memory and executed them in sequence

- Control logic to interpret instructions, plus input, output, and secondary storage

- A **stored-program fetch-execute cycle**: read instruction → decode → read operand → operate → next — the loop every processor still runs

- **Conditional branching**: an instruction could redirect control based on a computed result — the mechanism that makes loops and decisions possible

The draft also laid out an **instruction-set vision** — numbered operations, not just wiring — and treated the machine abstractly enough that it reads like a design for all computers, not one. It even included a neurological metaphor (organs, neurons) that helped and hindered the field in equal measure.

# Why "stored program" is the idea

Before the EDVAC report, a machine's sequence of operations was its *wiring* or its *card deck* — ENIAC's operators reprogrammed by hand-cabling panels for days. The EDVAC idea makes the program a **string of numbers in memory**: write it, store it, change it in software, run it. The machine can even modify its own program — which is what makes compilers, interpreters, and every layered software abstraction possible. The concept is so fundamental it no longer has a name in everyday talk; it is just "how computers work."

# The controversy

The report circulated under von Neumann's name alone — he was the one who wrote it up — but the ideas had been worked out jointly with **J. Presper Eckert**, **John Mauchly**, and the Moore School team. Eckert and Mauchly were furious; the uncredited publication cost them both the patent they wanted and, they felt, the credit. The Army judged the stored-program concept unpatentable because the report had already made it public. The resulting legal and personal rift shaped who got remembered: "von Neumann architecture" stuck as the name, though historians generally credit the design as a group product.

# Legacy

- The term **"von Neumann architecture"** descends from this document — though purists note its bottleneck (one bus, memory and code contending) is now called the *von Neumann bottleneck*.

- Its descendants ran within three years: the **Manchester Baby** first executed a stored program in **1948**, and **EDSAC** ran the first full working stored-program computer in **1949**. EDVAC itself — the machine the report described — was not finished until 1951, overtaken by the very designs it inspired.

- Every programming language, from [[FORTRAN]] to [[AdaLogic BASIC]], exists because instructions could finally live in memory and be treated as data.

# See also

- [[John von Neumann]]

- [[Manchester Baby]]

- [[EDSAC]]

- [[The first programmers]]

- [[History of Programming]]

# References

[[Category:History]]

[[Category:Hardware]]