Around 1950, after a number of modifications, the machine was set up in the Technical University of Switzerland in Zurich where it remained for several years, the only working computer in Europe . Today it is a historic model and can be seen in the Deutsches Museum in Munich. Unfortunately it's no longer in full working order. Another off-shoot of computer research ought to be mentioned here, too. By that I mean process control. At the Henschel Aircraft Factory Professor Herbert Wagner, for whom I worked as a stress analyst, was involved in developing remote- controlled bombs. To this end, the tailplane and wings - which were constructed of metal with a relatively low degree of precision - were subjected to detailed measurements using gauges at some 80 different points. The necessary adjustments were then calculated to allow for manufacturing inaccuracies. This required a rather complex calculation. Initially I constructed a special-purpose computer for a fixed sequence of operations using around 500 relays. This machine replaced a dozen calculators and worked very reliably for two years, two shifts a day. The procedure required a mechanic to read off the gauges. The values were then recorded and operators entered the figures into the computer. This led me to build an improved model which could read the gauges automatically and transfer data directly into the computer. The heart of the machine was a device that today would be called an analog/digital transformer. Perhaps this was the first process control system in the world. The machine itself had its own history. It completed trial tests on a production line in Sudetenland, but never reached full operational use as the War forced the whole factory to be re-located. The exact fate of the machine is unknown - it's possible the factory fell into the hands of the Russians, who must have been the only ones at the time to own a fully-operational computer. The Z3 had already been destroyed, the Z4 was not completed, and the first US machines, Mark I and ENIAC, were not operational at that time. However, it is unlikely that the Russians would have known what to do with the machine even if they had found it in an undamaged condition. Alongside my practical work with various computer models I started to consider certain theoretical aspects. The breakthrough to a new computing age went hand-in-hand with new scientific ideas and the development of new components. Today we talk about hardware and software. These expressions were really only introduced much later by the Americans although the terms have now become established. It was apparent that a special branch of "computer science" was needed. But shortage of time meant that I could only scratch the surface in this field. Initially I worked on my own, but towards the end of the War Herr Lohmeyer, an outstanding mathematician, was assigned to assist me. Lohmeyer was a product of Heinrich Scholz's school in Münster the latter himself a famous logician. The link with mathematical logic had already been established. As a civil engineer I was attracted by the prospect of drawing on predicate and relational calculus and exploring the possibilities they offered as a basis for computing. Take the frameworks used in building construction for example - were they not similar to the graphs used in relational calculus? Using pair lists, it was relatively easy to digitalize the structure of a framework with the aid of relational calculus, in other words, to break it down into its component data. This could then be entered into the combination memory, which had been invented by this time, and serve as a basis for combination calculations. This ought to mean that not only purely numerical calculations could be dealt with, but construction design itself. Up till now only the human mind had been capable of this. The same idea applied to frameworks and other types of building design. I became extremely pre-occupied with this new aspect of computing. I even went as far as learning to play chess in order to try to formulate the rules of the game in terms of logical calculus. Chess offered a mass of data structures within a limited space. A symbolic language (the expression "algorithmic language" was unknown to me at the time) which could describe chess problems seemed to me to be suitable for all computer machine problems. Plankalkül was later (1945) devised with this principle in mind.
This led to my first confrontation with what is known today as "artificial intelligence". Naturally I realized my computer would never be able to run that sort of calculation. But combination memory and the general circuitry were a step in the right direction. Many developments were predictable, of course, others were still in the realm of fantasy. I remember mentioning to friends back in 1938 that the world chess champion would be beaten by a computer in 50 years time. Today we know computers are not far from this goal.
But even in those days quite a lot was achieved at the drawing board. Today we call this computer architecture. The latter type of machine is known as the John-von-Neumann computer, after its namesake who first produced it 10 years later together with Goldstine and Burks. We now know this was a very elegant solution.
The question is why I did not use this concept in 1939 if I already knew about it. Well, at that time it would have been senseless to try to build that sort of machine, as the necessary facilities were simply not available. For example, storage capacity was not big enough to cope - an efficient program memory needs to be able to store several thousand words. Speed was also too low. It's true that floating point arithmetic can be performed by simply following a series of single instructions (as is the case today). But that means giving 10 to 20 times as many instructions. As long as the electronical prerequisites were not available it was a waste of time. Two things were needed first - high storage capacity - around 8,000 words, as in the first magnetic-drum memories - and electronic speed. Towards the end of the 40s this seemed possible, but as Germans we were not able to participate in this development at the time.
The possibility of a computer being able to deal with numeric calculations and logic organization was so exciting that I gave serious thought to a "logical computing machine". This led to the "program compiling machine" project. Work was to be split between numerical and logic computers. That included such areas as:
• construction of extensive programs made up of subprograms according to specific parameters
• address translation necessary for these programs
• capability of dealing with engineering structures (e.g. frameworks) using pair lists from which a numerical program could be developed.
This project was commissioned towards the end of the War by the Ministry of Aviation. However, the work soon proved to be too broadly-based and the ideas never left the drawing board. Since then, the concept of dividing computers according to numerical or logical operation has failed to find favor anywhere. The dominance of electronics made this unnecessary. The high speeds obtainable meant that such operations could be carried out by a single machine. However, one aspect became clear to me in view of all this research between 1936 and 1946. Some means was necessary by which the relationships involved in calculation operations could be precisely formulated. My answer was "Plankalkül" - today it would be termed an "algorithmic" language. However, in those days, known mathematical and logical forms were not advanced enough. There were also several other differences today's languages:
1. Plankalkül was not conceived as a means of programming the Z4 or other computers available at the time. Its true purpose was to assist in establishing consistent laws of circuitry, e.g. for floating point arithmetic, as well as in planning the sequence of instructions a computer would follow - what we would term "hardware" and "software" today.
2. It was meant to cover the whole spectrum of general calculating.
By contrast, the program languages which were developed around ten years later were relatively one-sided. They were designed specifically for existing computing machines, in other words, for the first really flexible electronic computers. In the first instance, these languages were concerned with conditional branching, address translation and suchlike. There was hardly any demand for logic21 operations, such as the application of predicate and relational calculus for engineering constructions, chess programs and so on. That also applied to the breakdown of data into yes-no combinations. In other words, mathematicians did not consider my principle of "data processing starting with the bit" to be of any fundamental importance. Plankalkül's weakness was that it went into too much depth with regard to difficult calculations which seemed better left to the future. The importance of my chess program, as an example of applied logic, was simply ignored. In addition to this, in the early fifties, I was completely absorbed in building up my business at a time when program languages started to become more relevant. This meant I could not participate in the debate on Algol and so on. Plankalkül was later published out of historical interest (in English, too, although it is now unfortunately out of print).
As has already been mentioned, 1945 was a hard time for the Germans. Our Z4 had been transported with incredible difficulty to the small Alpine village of Hinterstein in the Allgäu. My group and my Berlin firm had been dissolved. All of us who had managed to get out of Berlin were happy just to have survived the inferno there. Work on Plankalkül now continued in wonderful countryside, undisturbed by bomber attacks, telephone calls, visitors and so on. Within about a year we were able to set up a "revamped" Z4 in full working order in what had once been a stable. Unlike research going on in the USA, where every possible facility was available, our means were very limited.
There was no body or organization able to support our work at that time in Germany. Nevertheless word got out abroad that some sort of machine was operating in South Bavaria. IBM/USA instructed the German firm Hollerith GmbH to see what this was all about. All sorts of promising discussions ensued about possible applications for computers in various areas. But as everything was decided on the other side of the Atlantic at the time, no contract was signed. Interest was only shown in the industrial property rights. It wasn't even possible to secure a promise that I would be able to continue work on development. At that time computers were simply not that important.
However, we did have more success with Remington Rand, who commissioned us to continue development, initially in a special project dealing with mechanical relay technology. People didn't trust electronics fully at the time and wanted to have a second option. We ourselves were convinced that the future lay with electro-magnetics and electronics. This meant working against our own convictions. Nevertheless the machine was interesting as it became the first pipeline computer in the world - a punched card machine with a series of mechanical adding devices using mechanical circuits incorporated between the card reader and the card punch. Multiplication was carried out by a series of adding operations. This resulted in a reasonable degree of accuracy despite the relatively slow mechanics. But the development of electronic alternatives was progressing at that it was soon not worthwhile continuing with mechanical designs.
However, we were able to produce an interim solution based on our relays prior to the introduction of pure electronic machines, prototypes of which were being prepared by Remington in the USA. The production run consisted of around forty additional units for punched card machines, most of which were exported to Switzerland. About the same time we had been able to set up the Z4 as a scientific machine in the Technical University in Zurich. Both contracts helped us reestablish the ZUSE KG company, which I founded near Bad Hersfeld in 1949 with two other partners.
The German market slowly began to develop and we started to receive orders from German companies. One of our first clients was Leitz, which thanks to the world-famous "Leica" had the necessary financial means to purchase a computer for optics calculations. Later other optics firms followed suit, so that by the mid-50s we had a virtual monopoly in the field of scientific computers in the optics industry in central Europe.
By that time everyone was talking about electronic machines, but their reliability still left a lot to be desired. The first were built by scientific institutes for their own special purposes, as the commercially available machines were not good enough. We were able to bridge this gap with our relay computers. One thing in our favor was land consolidation, which was in full swing in Germany at the time. Computers were needed to calculate how fields and land were to be reallocated. To this end we developed the Z11 which was later further refined and used elsewhere in urban and agricultural surveying, optics, etc. This machine corresponded to what the users wanted - namely to always know how the calculation was progressing, and to have everything under complete control. In the course of time full automation led to a maze of branches and impenetrable procedures, and software became an increasingly urgent problem. We also experienced and influenced this, not always happy, development with our machines.
After some years, electronic components achieved a degree of reliability that warranted their production in large numbers. Initially we developed a tube model, the Z22, moving later to the transistorized versions Z23, Z31 and Z25. These were based on analytical code, which meant they were extremely flexible as far as programming was concerned. These were the first types of machine where general calculation using any desired data structure and program storage could be carried out. The machines were very popular with scientists and engineers, as they were excellent to play around with and all sorts of models for the most diverse problems could be tried out. Sadly, in most cases our machines could only serve to whet people's appetites. While our clients were very short of means during the early years of the electronic computer, research departments were later much better funded, allowing them to purchase much bigger, more expensive equipment. Unfortunately my firm hardly profited from this as we were oriented towards small and medium-sized companies. Nor did we have enough capital to take part in the development of larger machines. Our Z11, Z22 and Z23 are now sought after only as collector's pieces for museums and so on.
One development which received much impetus from specialists working in the surveying field was the automatic drafting board. The aim was the automatic representation of various maps which had been calculated beforehand by computer. It is interesting to note that the surveyors were looking for a very high degree of accuracy and initially only wanted to plot polygon vertices. From this emerged the Graphomat, the first computer-controlled automatic drawing board. Many others were interested, too. These were our first tentative steps in the direction of CAD. Little is known about this side of our work. Here, too, we found out that it is not always a good idea to be the early bird. As early as 1964 I was involved in negotiations with a major European carpet manufacturer to build a computerized control system for a large weaving machine I proposed that we start at the design stage of the carpet pattern. The intention was not to make the artisan redundant, but simply to give him a new tool. But this suggestion met with complete opposition from all parties concerned, and the contract failed to come off.
Competition in the computer sector became increasingly tough. Not only were the costs of hardware constantly rising, software development costs were also growing. My company with its thousand-odd employees faced growing capital shortages, making it necessary to bring in new shareholders. This led, step by step, to the company being completely taken over by Siemens.
Today, this leaves me free to devote more time to purely scientific problems, and I still work on a free-lance basis for Siemens AG Munich. I am currently involved in computer architecture, and am particularly interested in the parallel operation of machines. Back in the 50s I designed a machine for the meteorological office which today would be termed a "cellular computer". Here, too, however, I was guilty of trying to run before I could walk.
Perhaps I can round off by mentioning a few of the ideas I have examined on paper without ever being able to turn them into reality.
"The Computing Universe" is based on the idea that the whole cosmos is a kind of cellular computer, something that some physicists are seriously considering today. This theory has yet to be confirmed due to the lack of experimental evidence. A paper on the subject under the same title has also appeared in English[3]. I am sure the idea will gain considerable significance in the future and might help theoretical physics to solve a number of problems.
Another idea of mine was "The Self-Reproducing System". I approached this concept differently to John von Neumann, who dealt with it using pure mathematics in the context of cellular computers. As an engineer I was more interested in setting up the conditions necessary for actual construction. In essence, the idea envisages a tool factory which is capable of reproducing its own essential component parts. This idea has met with complete opposition. People have been reluctant to consider such a radical solution for all sorts of reasons. Today traditional means of production are being automated step by step. We have yet to build the factory of the future. But one day these far-sighted developments will become reality, leading to a complete revolution in the production process throughout the economy.
Of one thing I am sure - computer development has still a long way to go. Young people have got plenty of work ahead of them yet!
Biographical
Bauer, Friedrich L. "Between Zuse and Rutishauser - The Early Development of Digital Computing in Central Europe", in Metropolis, N., J. Howlett, and Gian-Carlo Rota. 1980. A History of Computing in the Twentieth Century, Academic Press, Inc., New York. pp. 505-524.
Ceruzzi, Paul E. 1981. "The Early Computers of Konrad Zuse, 1935 to 1945", Ann. Hist. Comp., Vol. 3, No. 3, pp. 241-262.
Czauderna, Karl-Heinz. 1979. "Konrad Zuse, der Weg zu seinem Computer Z3 (Konrad Zuse, the Path to His Z3 Computer)" Report 120, Gesellschaft für Mathematik und Datenverarbeitung, R. Oldenbourg, Munich, 105 pp.
Ritchie, David. 1986. The Computer Pioneers, Simon & Shuster, Inc., New York, Chapter 3.
Schwartz, H. R. 1981. "The Early Years of Computing in Switzerland", Ann. Hist. Comp., Vol. 3, No. 1, pp. 121-132.
Slater, Robert. 1987. Portraits in Silicon, MIT Press, Cambridge MA, Chapter 5.
Speiser, A. P. 1980. "The Relay Computer Z4", Ann. Hist. Comp., Vol. 2, No. 3, pp. 242-245.
Zemanek, Heinz. 1983. "Zuse, Konrad" in Ralston, Anthony, and Edwin D. Reilly, Jr. 1983. Encyclopedia of Computer Science and Engineering, Van Nostrand Reinhold Co., New York.
Zuse, Konrad. 1970. Der Computer - Mein Lebenswerk, Verlag Moderne Industrie, Munich.
Zuse, Konrad. 1980. "Some Remarks on the History of Computing in Germany", in Metropolis, N., J. Howlett, and Gian-Carlo Rota. 1980. A History of Computing in the Twentieth Century, Academic Press, Inc., New York. pp. 611-627.
Zuse, Konrad. 1980. "Installation of the German Computer Z4 in Zurich in 1950", Ann. Hist. Comp., Vol. 2, No. 3, pp. 239-241.
Copyright J. A. N. Lee, September 1994.