If you had asked to see 'a computer' before the outbreak of the Second World War, you might well have been taken to see a person who calculated numbers. Many businesses would use mechanical calculators operated by a hand crank; and the very largest may have used an electromechanical calculator in which information was stored on punched paper cards. The electronic computer was restricted to the pages of science fiction comics; but the foundations for the modern computer age were already laid.
As early as 1822, the British mathematician Charles Babbage had realised that the mechanical technologies of the Industrial Revolution could not only mass-produce physical items such as cloth or steel, but also process information. Babbage conceived his Difference Engine – a colossal construction of gears and switches that would generate mathematical tables with unparalleled precision. He never completed his Difference Engine (although a working replica now sits in the Science Museum), having devised a second revolutionary machine, the Analytical Engine.
Babbage’s Difference Engine was designed to complete a single task; but the Analytical Engine would perform any mathematical operation. In many ways it was the ancestor of all modern computers. Babbage’s machine would receive its instructions (what we would call a program) and information (data) on thousands of punched cards. Babbage’s steam-powered machine would have been the size of a house and composed of thousands of gears but its operation would have been remarkably similar to the modern computer!
The remainder of the 19th and early part of the 20th Centuries saw huge improvements in mechanical calculators, but crucially they were all restricted to performing a tiny range of tasks – such as adding, subtracting or multiplying numbers. None of these machines could be programmed to perform a new operation; if you owned a calculator that did not calculate fractions there was no way of adding that functionality – you would need to go and buy a new calculator that could process fractional values.
The computing revolution was unleashed in 1937 by the British mathematician Alan Turing. Turing was exploring the mathematical concept that there are certain logical questions that cannot be answered and was discussing ways in which the unanswerable questions could be identified.
Turing’s paper On Computable Numbers included a discussion of an imaginary machine capable of performing a simple mathematical task (say adding two numbers) by following precise, logical steps. Different Turing Machines would be built for performing different tasks. To perform a particular task, (say adding two numbers, multiplying the result by a third number and dividing that result by the fourth), a series of Turing Machines would be joined together in the necessary order – just like a child’s building blocks.
Turing then had an even more revolutionary idea. He conceived of a machine whose internal workings could be changed so that it could perform the same task as any of the simpler Turing Machines. Instead of a chain of simple machines, a single more complex Universal Turing Machine would perform all of the tasks, changing its configuration as instructed by commands punched into on a paper tape.
Turing’s research into computing was interrupted by the Second World War; he was recruited into Project Ultra – the top-secret attempt to break the German Enigma codes. Turing’s mathematical expertise produced a series of brilliant insights into the workings of Enigma, whilst his grasp of technology culminated in the bombe – an electromechanical machine that simulated the workings of Enigma and greatly speeded the task of decryption of coded messages.
As Europe lurched towards war in 1938, a novel electronic calculator was being built inside a Berlin apartment. Konrad Zuse’s ZI was the first calculating machine to use binary - zeroes and ones - to represent information. Instead of dealing with decimal numbers like humans, the Z1 performed calculations by the opening and closing of mechanical switches at very high speeds. The Z1 was not a true computer in that it could only be programmed with a limited range of instructions delivered on punched paper tape. The Z1 was damned by the limited precision of its mechanism, but it did illustrate the possibility of a programmable calculator. What Zuse needed was a more precise way to perform the complicated switching process.
Zuse’s Z2 replaced relatively slow and unreliable mechanical switches with much faster relays – electromagnetic switches used in telephone exchanges. By 1940 the Z2 was hard at work performing complex aerodynamic calculations for the German military. The same year Zuse founded Zuse Apparatebau – the first company dedicated to selling programmable machines and perhaps the ancestor of all computer manufacturers.
His Z3 was completed in 1941 and is today recognized as the first Universal Turing Machine. Despite this incredible breakthrough, Zuse was never afforded government backing and the Z3 was largely built from scrap components; its thousands of relays were those discarded by telephone exchanges, whilst data was punched into reels of discarded movie film. The Z3 could perform three to four additions per second, and took between three and five seconds for a multiplication.
Without knowing it, Zuse was in a race with the British. Project Ultra, based at Bletchley Park had made incredible breakthroughs against the German Enigma codes but was confounded by the code used to protect the most secure transmissions between German military commands. They named this mystery Tunny; to the Germans it was Lorenz.
By 1942 Ultra knew that Tunny was based on teletype – a technology for sending text messages over telephone lines at very high speeds using binary codes. The British even knew that Tunny encrypted messages using a technology known as the Vernam process which had been published before the war.
What the British did not know were the code settings used to transmit individual messages. Without this knowledge they resorted to attacking Tunny by hand – but the process would take weeks, by which time the intelligence itself was almost worthless.
Bletchley Park developed the Heath Robinson, an electro-mechanical machine that used two giant reels of paper tape to compare 1,000 encrypted characters against the possible Tunny settings every second. Heath Robinsons proved themselves to be extremely capable machines but the fragile paper tapes were prone to tearing and stretching as they raced through the mechanisms. A faster solution was needed.
Once again the solution came from the telephone industry; thermionic valves were widely used as switches in telephone exchanges. Unlike relays, valves could be switched almost instantly; in theory, a valve machine would be much faster than a mechanical computer. However, valves possessed only limited lifetimes; a computer would need thousands of fragile glass valves – could it perform any useful work before one of the crucial valves failed?
The Bletchley park mathematician Max Newman had developed a design for an entirely electronic code breaker, he sent these to Tommy Flowers, an engineer for the General Post Office, based at their research centre in Dollis Hill. Despite official scepticism, the world’s first completely programmable, electronic, digital computer was constructed in less than ten months.
Colossus I was delivered to Bletchley Park in December 1943 and immediately set to work cracking Tunny. It replaced one of the paper reels of the Heath Robinsons with an electronic representation in memory of the possible Tunny settings. Paper tape fed encrypted messages into Colossus at a rate of 5,000 characters per second where they were compared against the possible settings. Perhaps remarkably, a modern desktop computer is only slightly faster at cracking Tunny than its distant ancestor.
Colossus II arrived only six months later. Using 2,500 valves it was five times faster than its predecessor, (clearly computers going out of date as soon as they were built is not a modern problem), and proved so successful that a further nine machines were completed before the end of the war. Using Colossus the British regularly read Tunny messages before their opponents.
Despite early worries, valves proved to be extremely reliable. Colossus’ designers realised that most valve failures were caused when machines were turned on and off; therefore the Colossi ran almost uninterrupted from the time of its installation to the end of the war. Such was the prodigious amount of heat pouring from the machine that the Women’s Royal Naval Service operators were reduced to working in their underwear!
At the end of the war, on the express orders of Winston Churchill, eight of the Colossus computers were reduced to scrap. Every part of Ultra was classified and Tommy Flowers was ordered to burn the original plans. He complied and Colossus vanished from sight for almost thirty years. The two remaining Colossus machines were rebuilt at GCHQ near Cheltenham where they remained in service until 1960.
Such was the secrecy surrounding Ultra and Colossus that their very existence only became public knowledge in 1974. Until that time, it was widely believed that the first computer was 1945’s ENIAC, a colossal machine designed for the United States Army.
Zuse’s Z3 was destroyed during an air raid, but the partially completed Z4 not only survived the war but was sold to a Swiss university becoming the first computer in continental Europe. Zuse’s company went on to sell hundreds of machines before becoming part of the Siemens computer business.
After World War II, Alan Turing continued to be a brilliant researcher; his ACE was one of the first public computers in Britain, he programmed the first game of computer chess, made key discoveries in the field of artificial intelligence and described the mathematics underlying patterns in biology. Tragically, he committed suicide in 1952, aged only 41.