Carel S. Scholten and Bram Jan Loopstra
Scholten accepted the job solely based on the knowledge received from the following (translated) telegram while being on vacation "Do you wish to participate in the building of an automatic computing machine. Mister Loopstra has already agreed to. Sender: Mathematical Center." The terms "Automatic computing machine" and "Mathematical Center" were new and interesting concepts. Because he had no money for a telegram he replied with a postcard saying no more than "Agreed," and Scholten continued his vacation
Scholten and Loopstra met each other in high school after which they studied physics together. They were still close friends when in 1946 they were hired as the first two members of a new department at the Mathematical Center that was to build a computer. They didn't really have any relevant knowledge to start with though, not from their physics study or otherwise.
Edsger Dijkstra, Bram Loopstra and Ria Debets
Differential Analyzer
The first place they could work in was a room in the National Physics Laboratory where professor Clay was in charge. Their first action was to demolish the high voltage installation that was occupying the room, this to the dismay of professor Clay as he was very fond of it, but he wasn't present and came too late. Then they decided it would be useful to equip the room with 220V sockets, so they went to the Waterlooplein to buy a secondhand hammer, pincers and screwdriver, some wire and some wooden sockets (it was 1947 and still a lot of after-war shortages). For a moment they were debating if they could reasonably submit the, in their eyes, exorbitantly high bill, but they did so nevertheless.
From that moment they weren't sure what they should do next, so they went to consult the scarce literature about the subject. They could find only two pieces. The first was about the ENIAC, a digital (decimal) computer, and one about a differential analyzer, a device intended to solve differential equations with continual variables represented by changing physical sizes, in this case the turning of axes. Because the piece about the ENIAC was badly written and used fearsome numbers like needing 18000 tubes, they decided to try and build a differential analyzer.
Either of the two things, rebuilding ENIAC or building a differential analyzer, were not within their abilities as they lacked training in electronics, and they had only seen one small practical example with a radio tube, during a student assistant job where they were supervising a physics practical for doctors.
Hence they started with trying to figure out how the tubes worked. Their first test could come directly from a slapstick movie as they were taking cover behind a table, laid down sideways, seeing if they could find out how the triode worked. The only difference with a slapstick movie was that nothing happened.
With the aid of some textbooks and 'tube books' from some manufactures they brought their electronics knowledge up to such a level that they succeeded in bringing a pair of components to a predictable state.
Carel Scholten, here soldering the ARRA I
Engineering the ARRA
From January 1948 Scholten had to serve in the military. When he got back in September 1950 the group had moved to a new place, half of a school building on 49 2nd Boerhaavestraat. The department had also expanded and they were now with five. The most important change was in the plans they had though. The idea of the differential analyzer had been abandoned as it had become clear that the future was in digital computers, and a part of such a digital computer had already been realized: the ARRA (I).
Frequently they traveled to Eindhoven to visit Philips. On the way back they mused about the outstanding lunch they got there and then started to look if the catch was worth the trip. In most cases it was. It made the ARRA I a sample sheet of what kind of electronic components were available in the industry.
When in November 1952 Gerrit Blaauw joined their group they reacted carefully, as Scholten put it: "We weren't very obedient boys. We were aware that we didn't know it all, but we considered it highly unlikely that somebody else would know better. Therefore we looked at the newcomer with suspicion. All the greater was the achievement of Gerrit that his lecture convinced us of the sensibility of his proposals." And that were some serious changes, he proposed to make a clocked machine, uniform buildings blocks, pluggable and therefore changeable units, a neat design method and decent documentation.
Designing the ARRA II
To Scholten, the design phase of the ARRA II marks the point where designing computers started to become a trade. Mostly because of the introduction of uniform building blocks and that the machines could be described in a multidimensional binary state space that made it useful to use tools like boolean algebra. This is also why when Scholten designed the signed addition multiplier for integers ((A,S) := (M) * (S) + (A) for all sign combinations of A, S and M) for the ARRA II. It would be the first time he could remember to have dedicated a document to the proof that the suggested solution was correct. Of course the proof was in a form that wouldn't suffice today, but still.
"It indeed worked as intended, and you can imagine my amusement when after several years, from a French book about computers, I was informed that this problem was insolvable."
After building a near copy of the ARRA II, the FERTA for Fokker they started designing and building the ARMAC which took only about 18 months.
They saw the end of the amateurism approach with the construction of the ARMAC. Computers started to become available on the market and it started to be painfully clear that their group didn't belong in the Mathematical Center. A gradual shutdown would be the destruction of good know-how they had, thus they looked for a solution. The Mathematical Center found this with the Nillmij - a Dutch insurance company - and together they started the Dutch computer industry in the form of Electrologica. The employees were gradually transferred from the Mathematical Center to Electrologica. Scholten officially transferred at January 1st, 1959, but he had already been building the X1. Its prototype was making its first calculations at the end of 1957.
Retrospect
When Scholten looks back at the early days he sees that "Designing computers" didn't actually exist. Mainly because the activities that could be described as such went down below the abundance of other concerns that needed their attention. Those who were involved in the construction of big computers worked mostly in very small team and did all the work that had to be done.
Scholten and Loopstra decided about the construction of racks, placement of doors and fastenings, placing of ventilators (the ARMAC used 10kW), they set up high capacity power distribution cases and the cables that went with that, knew all the available fuses and diameters of electrical wires by heart, they soldered, peered at oscillators and climbed the machine armed with a vacuum cleaner to clean it. Occasionally they also did a bit of designing.
It shouldn't be idealized, the technical problems brought them to the edge of despair every now and then. Inadequate components caused problems, and even so much as the lack of knowledge and insight. And this wasn't something special of the group at the Mathematical Center, the whole world hadn't mastered the trade yet. It was a fascinating time, continually knowing that they were working on something that hadn't been done before (or at least not as far as they knew). It was a time in which the arranging of working overtime, if need be even whole nights long, didn't cause any problems. And it was a time in which they knew most of the participants of the international conferences about computers, at least by sight[1][2].
References
- Zij Mogen Uiteraard Daarbij De Zuivere Wiskunde Niet Verwaarloozen, , Amsterdam, Centrum voor Wiskunde en Informatica (CWI), 1987 .
- Computers Ontwerpen, Toen, , 1979 .