We Just Found the User Manual for the First Digital Computer Ever Built
These days, losing the manual for some piece of electronics you’ve purchased is notable mostly because you had a printed document to lose in the first place. In the dead-tree dominated days of yore, of course, this was less true. Documentation loss is a major problem in the effort to understand old computer systems, and it’s part of what drives ongoing data preservation efforts across the industry. Until recently, the Zuse Z4 could have been a poster child for this sort of problem.
The Z4 was the brainchild of Konrad Zuse, a German who deserves to be better known than he is for his early, groundbreaking work. Zuse had the misfortune to be making some of his biggest breakthroughs immediately prior to and during World War II. It was Zuse who designed the first high-level programming language from 1942 to 1945. This is remarkable because, as Wikipedia notes, Zuse had no training whatsoever in mechanical computing devices. He independently discovered both propositional calculus and lattice theory, calling them “combinatorics of conditionals” and “study of intervals,” respectively.
The Zuse Z4 is the oldest preserved digital computer in the world and arguably* the first digital computer. The Z4 was developed through the end of the war and was moved multiple times while under construction to keep it away from the advancing Soviet army. After the war, it was expanded and became the second digital computer in the world to be sold. The preserved model is on display at the Deutsches Museum in Munich and is pictured above.
Its documentation, however, was a different story. A recent blog post by the Association of Computing Machinery details how the rare documents were found. Archivist Evelyn Boesch, with ETH Zurich University, contacted Herbert Bruder of the ACM and informed him that her father, René Boesch, had kept a tranche of rare historical documents. These turned out to include a user manual for the Z4 Zuse, as well as notes on flutter calculations. Other documents, dated October 27, 1953, detail what the Z4 was working on. At the time, it was being used to perform flutter calculations on the Swiss FFA P-16 fighter aircraft, which was then in development. Details from the recovered documents show that it took the Z4 50 hours to simulate 2.4 seconds of flight time, which is slightly worse than the current version of Microsoft Flight Simulator.
The ACM blog post notes that “around 100 jobs were carried out with the Z4 between 1950 and 1955,” implying an average per-job computation time of about three weeks.
What We Learn From Manuals Like This
The recovered Z4 manual illustrates why this type of document preservation is so important. From their earliest days, computers were upgradeable — machines like ENIAC were outfitted with the equivalent of RAM upgrades and CPU improvements. In the Z4’s case, support for conditional jump instructions was added post-manufacture. The only problem was, nobody could remember exactly how the feature worked. ACM notes: “However, in a survey a few years ago, the few surviving eyewitnesses could not remember how it was executed.”
Page 8 of the manual provides this information. My German is rusty, my technical German is nonexistent, and frankly, the images are a bit tough to read, so I’m not going to try to translate exactly how the function worked. Without information like this, it would be impossible to precisely replicate or understand how the Z4 embodied or improved upon the computational capabilities of the time.
*The answer to “Who invented the first computer?” is essentially arbitrary and depends entirely on how you choose to define the term “computer.” The UK’s Colossus is declared the world’s first “programmable, electronic, digital computer,” by Wikipedia, but it was programmed by switches and plugs, not a stored program. The Z4 is considered to be the first commercial digital computer but it’s not electronic. The first electronic stored-program computer is the Manchester Baby, but Konrad Zuse’s earlier Z3 could store programs on tape — it just wasn’t electronic. Other obscure machines, like the Atanasoff-Berry Computer, were not Turing-complete and couldn’t store programs, but still contributed critical ideas to the development of computing.
Also, if you were taught that ENIAC was the first computer (or digital computer, or electronic digital computer, etc, ad nauseam), that’s more propaganda than fact. ENIAC was more directly based on machines like Colossus than was known at the time, because the wartime efforts of the British remained classified, while ENIAC was widely celebrated in the media.
Finally, reading up on the history of early computing is a good reminder of how many people, institutions, and companies contributed various technologies and principles to the field. One reason you can subdivide the question of “Who built the first computer” to such a fine degree is that there were so many “firsts” for someone to achieve. There was a time in the 1930s and 1940s when mechanical, electromechanical, and digital systems were sharing space and serious development dollars simultaneously. We don’t have anything remotely equivalent today, and even our wildest architectural departures from the x86 “norm” are still based on digital computing. That could change in the future, if Intel’s MESO architecture comes to fruition and proves capable of replacing CMOS in the long term.
But for now, the 1930s and 1940s represent a tremendously dynamic period in computing history that we don’t really have an equivalent for — though some of the quantum computing work is getting really interesting.