Zuse, Konrad

views updated

ZUSE, KONRAD

(b. Berlin, Germany, 22 June 1910;

d. Hünfeld, Germany, 18 December 1995), logic, computers, programming, computer industry.

Zuse is popularly recognized in Germany as the “father of the computer,” having built the world’s first programmable computing machine in 1941. Zuse is less well known in other countries because most of his early computers were built during World War II and became famous in and outside Germany only several years after the war.

Early Years. Konrad Zuse was born in Berlin to Emil and Maria Crohn Zuse. His father was a Prussian civil servant working for the postal service who relocated the family to Braunsberg (now Braniewo in Poland) when Konrad was still a child. Konrad attended elementary school in that town and began studying at the local Gymnasium Hosianum. The family moved again in 1923 to Hoyerswerda (a town in Germany near what is now the border with Poland). In Hoyerswerda, Zuse was registered at the Realschule, a school that allowed pupils to continue studying at any of the several technical universities established in Germany. The family eventually moved back to Berlin and Konrad Zuse began his studies at the Technische Hochschule Charlottenburg (renamed Technical University of Berlin after World War II). Zuse started studying mechanical engineering, changed to architecture, thought for some time of becoming a commercial graphic designer, and settled finally on civil engineering. Years later Zuse wrote in his autobiography that he eventually discovered civil engineering to be the ideal field for him because he could combine his artistic interests with his technical prowess, especially regarding mechanical constructions. The young Konrad Zuse was an inventor and a tinkerer, often withdrawing to work with his “Stabil” mechanical set (a German version of the Meccano or Erector Set). As a student he won several prizes for his constructions, which he enjoyed showing off.

As part of his civil engineering studies at the Technische Hochschule, Zuse learned to perform repetitive static calculations like those needed to determine the stress on materials of structures such as bridges or cranes. Static calculations were performed completely by hand or with the help of desk calculators. Spreadsheets, on which all necessary formulas had been preprinted, were laboriously filled row by row. It was tedious and repetitive work that led Zuse to consider the possibility of automating the task. If engineers simply had to fill in data and follow a fixed computational path, then a machine could take over.

The Mechanical Programmable Machine. After his graduation in 1935 Zuse started working as a stress analyzer for the airplane manufacturer Henschel Flugzeugwerke. He kept this position for less than a year, resigning with the purpose of starting his own company. He wanted to build automatic calculating machines and had already made contact with Kurt Pannke, a constructor of mechanical desk calculators. However, Zuse’s short-lived employment at Henschel would prove crucial for him in later years: Twice in his life his superiors at Henschel would help him secure a deferment from the army, both times arguing that he was needed as an engineer and not as a soldier on the battlefield.

In 1936, with his parents’ financial support, Zuse began to build the automaton that so far had only existed in his imagination. Some friends at the university assisted by working for him while others offered small monetary contributions so that he could finish what would become the machine V1 (Versuchsmodell 1, “experimental model one”). Perhaps the most important difference between Zuse and other computer inventors working in the late 1930s was that Zuse was designing his machine essentially alone, whereas in the United States scientists such as John Atanasoff and Howard Aiken had the resources of universities or important companies at their disposal. The entire mechanical conception of the V1 (later renamed to Z1) was his brainchild.

Zuse, ignorant of the internal structure of any type of calculator built at the time, started from scratch and developed an entirely new kind of mechanical assembly. Whereas contemporary desktop calculators were based on the decimal system and used rotating mechanical components, Zuse decided to use the binary system and metallic shafts that could move only in one direction. That is, the shafts could only slide from position 0 to position 1, and vice versa. Such shafts were all that was needed for a binary machine, but important obstacles had yet to be surmounted. It was necessary to design the complete logical description of the machine and then “wire” it accordingly. The mechanical components, however, posed a formidable challenge because every movement of one logical gate had to be mechanically coupled with the movement of the other gates. Horizontal displacements of the components had to be transformed into sliding displacements across different layers of the machine, or even into vertical displacements. From an early twenty-first century perspective, the mechanical design of the machine was much harder than conceiving the pure logical structure. It is fair to say that none of Zuse’s friends understood exactly how the machine worked, although they spent weeks manufacturing the hundreds of metallic shafts needed for the apparatus.

The Z1 was operational in 1938. It was shown to several people who saw it rattle and compute the determinant of a three-by-three matrix. The machine, however, was not reliable enough. The mechanical components, all machined at home, had a tendency to get stuck. Nevertheless, the mechanical Z1 proved that the logical design was sound. Therefore, an electrical realization, using telephone relays, could be contemplated as the next step. Helmut Schreyer, an electronics engineer and college friend of Zuse, suggested the use of vacuum tubes. Schreyer, in fact, adopted this as his PhD project and developed some vacuum tube circuits for an electronic machine. Zuse, however, was not convinced that vacuum tubes should be used, although they promised extremely fast calculations. He doubted that in the long run vacuum-tube machines could be made to perform as reliably as relays or even mechanical components. Zuse had already been contemplating possible uses for his machine: His goal was the development of a programmable replacement for

mechanical desktop calculators for deployment in large or medium-size companies. This was to be a “computing machine for the engineer,” eventually so small that it could be placed on top of a desk.

In 1938 Schreyer and Zuse explained some of the electronic circuits to a small group at the Technische Hochschule. When asked how many vacuum tubes would be needed for a computing machine, they replied that two thousand tubes and several thousand other components would be enough. The academic audience was in disbelief: The most complex vacuum circuits at the time contained no more than some hundred tubes, and the electric power necessary to keep such a machine working would be prohibitive. Just six years later the ENIAC, built at the Moore School of Electrical Engineering in Philadelphia, would show the world that vacuum-tube machines were indeed expensive but entirely feasible.

The start of World War II had immediate consequences for Zuse; he was called to serve in the army, and was for six months deployed on the Eastern front. With the help of Kurt Pannke, Zuse tried to obtain a transfer to Berlin in order to continue his work on the next computing machine. Helmut Schreyer, who worked as an engineer at the university, also tried to obtain Zuse’s discharge by offering to build the military an automatic air-defense machine that could be operational in two years. His offer was met with the sardonic reply that the war would be over by then. Finally Zuse’s previous superiors at Henschel were able to obtain his transfer to the Henschel airplane factory in Berlin-Adlershof, where he was hired to make the calculations necessary to correct the wings of the “flying bombs” (now called cruise missiles) being built in Berlin.

In 1940 Zuse started working for the Special Section F at the Henschel factory. During the next five years he developed the machines S1 and S2. The latter could automatically measure some parameters of missile wings, transform the analog measurement into a digital number, and compute a correction to the wing based on these values. The earlier model, the S1, needed such numbers to be typed on a decimal keyboard. The S1 and S2 were probably the first digital computing machines used for factory process control. The measurement instrument used in the S2 was also almost certainly the first industrial analog-to-digital converter, although it was never used in real production. Both machines were, from the computational point of view, subsets of the machines described below. Their existence remained unknown to the public at large for many years after the war.

In 1940 Zuse put together the machine Z2, an experimental model that used an integer processor built out of relays and a mechanical memory cannibalized from the Z1. This machine helped Zuse convince the German Airspace Research Office (DLV in German) to partially finance the development of the successor to the Z1, the Z3, which would be built using only relays. The Z3 became operational in 1941. It had the same logical design of the Z1 but was built with electrical telephone relays.

Structure and Capabilities of the Z1 and Z3. The Z1 and Z3 worked with floating-point numbers (that is, numbers such as, for example, +12.654 with an integer and a fractional part). Zuse developed an internal numerical representation that strongly resembles the internal number format used in modern computers. Each number was stored separated in three parts: the sign of the number, the exponent of the number in twos complement notation, and the mantissa of the number. In order to handle each part, the processor of the Z1 and Z3 consisted of two main blocks, one for processing the exponents of numbers and one for processing the mantissas.

The two machines, Z1 and Z3, shared a common architecture. Their main components were:

the memory for storing numbers (sixty-four in total);
the processor for computing;
A punched tape for storing the sequence of program instructions; and
an input-output console.

The instructions were read from the tape and were executed one by one by the processor. The console allowed the user to enter decimal numbers with a decimal keyboard (similar to the keyboard of a cash register) while the results were shown in a panel with digits illuminated by lamps.

The instruction set of the Z1 and Z3 consisted of the four arithmetical operations (addition, subtraction, multiplication, and division) as well as the square root operation. There were four additional operations for reading and displaying results and for moving numbers between processor and memory. The Z3 was very much like an early electronic calculator of the 1970s but much slower; a multiplication required eighteen machine cycles and was executed in three seconds.

Using the instruction set mentioned above, it was possible to process any arithmetical formula of the kind used in engineering applications. However, the instruction set did not provide a conditional branching instruction, so that it was relatively difficult, although not unfeasible, to perform more complex computations. Also, the two ends of the punched tape could be bound to form a loop, so that repeated execution of the same program was possible.

Zuse avoided the use of an excessive number of logical gates for the processor by relying on control units that worked as microsequencers, one for each command in the instruction set. A microsequencer consisted of a rotating arm that advanced one step in each cycle of the machine like a rotary dial. A clock (a rotating motor) provided the clock cycles needed to synchronize the machine. In the case of the Z3, the operating frequency was set at five cycles per second. Five times per second the rotating arm in a microsequencer activated the next step of the operation at hand. For example, in the case of multiplication, repeated addition and shifting of numbers were needed (as happens when two numbers are multiplied by hand). The eighteen partial operations needed were all started by a microsequencer with eighteen contacts for the rotary dial. The microsequencer can thus be thought of as a kind of hardwired program that reduced very complex instructions to a sequence of simple operations. Therefore, modifying the complete internal operation of the machine consisted only of rewiring the microsequencers without having to modify the rest of the processor. This resulted in a very efficient and flexible architecture, explaining how Konrad Zuse was able to build a machine that rivaled the British or American computers built during the same period, even with only a hundredth of the resources at his disposal.

During World War II, Zuse worked continuously for the Henschel factory but was able to start his own business in 1941. The Zuse Ingenieurbüro und Apparatebau, Berlin, was the first company founded with the sole purpose of developing computers. The Z3’s successful demonstration brought Zuse a contract with the German aircraft research unit (DLV) to develop a still larger computer, the Z4. This machine had a very similar design to the Z3 but would have 1,024 memory words instead of only 64. The machine was built and was almost operational by early 1945, when Russian troops approached Berlin.

War’s Aftermath and the Plankalkül. Zuse fled with the Z4 before Berlin fell to the victorious Soviet army. One of his collaborators was able to obtain train transport for the machine, somehow managing to smuggle it as a valuable military asset. The Z1 and Z3 had already been destroyed by air raids during the war, so that the Z4 constituted the only asset of Zuse’s company. After several detours Zuse established himself in Bavaria, where he would survive the following years by painting, consulting, and attempting to restart his company. During this period of forced inactivity, he finished his manuscript on the Plankalkül, a remarkable document first published in the 1970s.

The Plankalkül (calculus of programs) was the first high-level programming language conceived in the world. It was designed by Zuse between 1943 and 1945, that is, at a time when the first computers were being built in the United States, United Kingdom, and Germany. It represents one of the major achievements in the history of ideas in the computer field, although it was first implemented in 1999 by a team of researchers in Berlin.

The Plankalkül corresponded to Zuse’s mature conception of how to build a computer and how to allocate the total computing work to the hardware and software of a machine. Zuse called the first computers he constructed “algebraic machines” in contrast to “logistic machines.” The former were specially built to handle scientific computations while the latter could deal with both scientific and symbolic processing. Zuse’s “logistic machine” was never built, but its design called for a one-bit word memory and a processor that could compute only the basic logic operations (conjunction, disjunction, and negation). It was a minimalistic computer in which the memory consisted of a long chain of bits, which could be grouped in any desired form to represent numbers, characters, arrays, and so on. In some ways the logistic machine resembles Alan Turing’s proposal of 1936, later known as the Turing Machine.

The Plankalkül was the software counterpart of the logistic machine. Complex structures could be built from elementary ones, the simplest being a single bit. Also, sequences of instructions could be grouped into subroutines and functions so that the user dealt only with a powerful high-level instruction set that masked the complexity of the underlying hardware. The Plankalkül heavily exploited the concept of modularity, which later became so important in computer science: Several layers of software made the hardware invisible for the programmer. The hardware itself was to be simple and only able to execute the minimal instruction set.

In Plankalkül the programmer uses variables to perform computations. There are no separate variable declarations: Any variable can be used in any part of the program, and its type is written together with the name. Variable assignment is done as in modern programming languages where a new value overwrites the old value. Many operations are those used in modern programming languages (addition, subtraction, and so on).

Plankalkül is universal. It can deal with conditional instructions of the “if then else” type and makes available an iteration operator W that repeats the execution of a sequence of instructions until a loop-breaking condition is met. Using these constructs, any kind of computation can be coded with Plankalkül.

Although Zuse published some minor papers about the Plankalkül and tried to make it known in Germany, the language fell into oblivion. The main problems were its ambitious scope, the large variety of instructions that it contained, a modular architecture that called for incremental compilation, and the presence of dynamical structures and functionals. Some aspects of the definition were not quite clean, and the absence of type checking would have made it extremely difficult to debug. A practical implementation of the Plankalkül certainly requires a major revision of Zuse’s draft of 1945. However, Plankalkül was very much ahead of its time considering that many of the concepts on which it was based were rediscovered much later. It would take many more years for programming languages to achieve Plankalkül’s level of sophistication.

Rebirth of Zuse’s Company. After World War II, Zuse’s company was revitalized when Professor Eduard Stiefel, from the Technical University of Zurich (ETH), drove to Bavaria to see the refurbished Z4 in operation. He decided to rent the machine for his university. The Z4 was installed in Zurich in 1950, several months before the first UNIVAC was delivered in the United States, and was therefore the first commercial computer in operation in the world. For several years the Z4 was the only computer operating in continental Europe. The machine had the same logical structure as the Z3 but contained more memory and an expanded instruction set. It was used for many years at the ETH and is now part of the history of computing exhibition of Deutsches Museum in Munich. It is the sole Zuse machine built before 1945 that has been preserved.

Zuse’s company (with the new name Zuse KG) flourished after the war, and many other machines were built. They were all numbered progressively (e.g., Z5, Z11) according to their introduction. For some years Zuse continued building relay computers and even argued in favor of micromechanical elements. Gradually, however, the electronic components were miniaturized, their reliability increased, and with the dominance of American companies in this field, Zuse KG had no choice but to develop vacuum-tube and transistor-based machines. The first Zuse KG transistorized computer was the Z23, a commercial success: Eighty machines were delivered in Germany and eighteen to other countries. The German Research Foundation actively promoted the machine and subsidized its introduction in universities, where it was used to jump-start most of the computer science education in universities.

The Z23 and the Z22 (built with vacuum tubes) were remarkable in that they constituted the first radical departure from the architecture of all previous Zuse machines. Their internal structure consisted of serial registers, which allowed the use of fewer components. The number of instructions was kept to a minimum. A compiler allowed programmers to write code with a syntax that was in between assembly code and a high-level programming language. After the Z22 and Z23, Zuse would often confide that the new machines were being designed not by him but by his engineers.

Another important development, and Zuse’s last encore, was the introduction of the Graphomat in 1961, a plotter that could be used by architects and geologists to generate diagrams and drawings. The Graphomat could be connected to the Zuse computers and used gears that provided smooth, continuous movement in each direction. The gears were designed by Zuse himself.

The Z23 and the Graphomat were successful, but the development of the next line of computers proved too costly. Eventually the dominance of the U.S. computer industry in Europe, as well as the late adoption of a fully electronic design, brought financial difficulties to Zuse KG. The company was sold first to Brown Boveri and Company in 1962 and later to Siemens. Production of the Zuse series of computers was eventually stopped. Zuse retired after the Siemens takeover and received retirement benefits. In the ensuing years he continued writing, applying for patents, and making a case for his place in the history of computing.

In retrospect it can be said that Konrad Zuse’s greatest achievement was the development of a family of fully digital, floating-point, programmable machines that were built in almost total intellectual isolation from 1936 to 1945. His dream was to create the small computer for business and scientific applications. He worked single-mindedly during many years to achieve this objective. His 1941 patent application for the computing machine Z3 was refused in 1967 by a German judge as it was deemed to lack “inventiveness.” The decision on the application was delayed so long, firstly, because of the war, and secondly, because a number of major computer companies battled against Zuse in court. Zuse, however, always considered himself the one and true inventor of the computer, and his public statements on this subject demonstrated some bitterness about his lack of recognition in other countries.

Epilogue. Konrad Zuse married Gisela Brandes on 6 January 1945. Gisela gave birth to their first son a few months later, and four more children followed in the ensuing years. But Konrad Zuse was not a family man: Over the years his sole obsession was starting and leading his company. After his retirement he was much decorated in Germany, receiving, among other distinctions, the Federal Cross of Merit and the Siemens Ring. He was named a fellow of the Computer History Museum in California in 1999. Several honorary doctorates, as well as a professorship, were bestowed on him. Furthermore, the most important prize in Germany in the field of computer science bears Konrad Zuse’s name. Zuse died on 18 December 1995, at the age of eighty-five.

His early machines have been reconstructed: A model of the Z1 was built in the 1980s by Zuse himself and is on display at the German Technology Museum in Berlin. The Z3 was reconstructed by Zuse’s engineers in the 1960s and is on display at Deutsches Museum in Munich. A new functional replica of the Z3 was built in Berlin and is on display at the Zuse Museum in Hünfeld, Germany, where several of Zuse KG’s computers are also housed.

It has been frequently said and written that the computer is a by-product of World War II, or at least that its birth was catalyzed by the events surrounding that conflagration. In the case of Konrad Zuse this is only partially true. The inspiration for his first computing machine, the Z1, predates the war. The six months that Zuse spent on the Eastern front in 1939–1940 were certainly an interruption of the project he already had been working on for almost three years. If the war had not started, the Z3 computing machine would have been built sooner. But once hostilities broke out, Zuse at least was able to convince the military establishment that computing machines were useful for aerodynamical numerical calculations. The successful demonstration of the Z2 prototype led to a contract with the German Airspace Research Office (DLV), which financed most of the construction of the Z3. Once the Z3 was operational Zuse built the special-purpose machine S1 and also started building the more powerful computing machine he had being dreaming about all those years, the Z4. The construction of the Z4 was done under a war contract financed by the German military until 1945.

Although almost no one in Germany fully understood the importance of Zuse’s work, at least the people in charge of the strategic management of aeronautic research and development recognized the relevance of fast computations. It is noteworthy that Zuse could leave the Eastern front and be freed of day-to-day responsibilities at the Henschel Werke in order to attend to his own company. This would not have happened if the military experts had not thought that his company was useful and necessary for the war effort.

Konrad Zuse was no resistance hero, but he certainly never tried to gain office or position himself in academic politics. While professors and researchers at German universities, especially at the Technische Hochschule Charlottenburg, flocked to the Nazi party in order to advance in their professions, Zuse’s own career was cut short by the war. Unfortunately, not much is known about his political views at the time. In his memoirs Zuse deals with the regime and politics during the war in just a few paragraphs. Ideologically he was very much impressed by Oswald Spengler’s theory of the decline of Western civilization. He continued to mention Spengler in his late years.

It was probably Konrad Zuse’s personal tragedy that he conceived all the elements of the computer sooner and more elegantly than any other computer pioneer but was living in Germany when the country was on the path to self-destruction. Outside of Germany, and outside a very small circle for that matter, nobody took notice of the Z1, Z2, Z3, and Z4. The S1 and S2 were secret machines. Zuse’s work was not rediscovered until the late 1940s, and by then it was too late for his machines to have had any serious impact on the design and construction of modern computers. Zuse’s work was worth a footnote, at most, in early scholarly books about the history of computing. This has changed since the 1990s, as more has become known about the life and work of this most remarkable computer pioneer.