Electric Universe

by David Bodanis

  In late 1940 this was one of the most vital of all British operations, for if more of the U-boats that were destroying Atlantic convoys could not be found, Britain would slowly but inevitably starve. His unit needed to build some sort of device that could at least partly duplicate the inner workings of the German machines and churn through tens of thousands of permutations each hour. The British device, running with cogs and punch tape and simple circuitry, was given the label “bombe.”

  It wasn’t a computer, for lots of clerks were needed to run the bombe devices, first dozens of them, then hundreds. Almost all were young women, recruited usually from proper upper-middle-class families into the Women’s Royal Naval Service—Wrens, for short. But there was something curious about this combination of humans plus machines. For with the Wrens and the bombe device working together, Turing had created the closest thing yet to the universal computer he had imagined in the quiet prewar years. That combination of “Wrens plus the simple electric circuits in the machines they supervised” was, in a sense, the “hardware.” When fresh data came in from the outside world—when the Germans switched to a different encoding procedure, or a colleague proposed a new approach—Turing simply had to “reprogram” the whole group working for him.

  Progress was slow at first. Sometimes convoys sailed toward areas where the Admiralty knew there were packs of U-boats waiting, but the clacking Wren-directed machinery couldn’t break the intercepted German shortwave signals in time. They would realize with dread that ships were exploding, sailors drowning. But Turing was a master at organizing small groups, and increasingly often they succeeded. His “computer” didn’t yet exist as one centralized physical object, but he now had the equivalent of most of its parts—the memory, the processor, the reconfigurable software—working away, albeit scattered in separate buildings, and made of such disparate elements as female Wrens, copper electrical wires, and Turing’s own thoughts.

  And then he fell in love—or at least very much “in like.” Joan Clarke was a young Cambridge mathematics student who’d been sent to his division at Bletchley. He told her that he had “homosexual tendencies,” but she didn’t seem to mind, and they began the sort of romance that any overworked, mathematically fascinated young couple would find normal. In their free moments they would lie on the Bletchley Park lawn together, commenting on the mathematical patterns in the daisies they plucked for each other. At dawn, after exhausting nine-hour night shifts, they played bouts of what Turing called “sleepy chess” (using clay pieces he baked on the hot-water radiator in his lodgings). Once Turing knitted a pair of gloves, but wasn’t dexterous enough to finish off the fingertips; Joan, only slightly teasingly, did it for him.

  The greatest electrical progress began in February 1942, when the German navy shifted to new forms of coding. All the advance intelligence that Bletchley had been providing on U-boat wolf packs suddenly disappeared. Yet airborne radar wasn’t quite ready to take up the slack. The Royal Navy, and its new American allies, were operating blind. Turing was tense; he picked at the sides of his fingers so much that an open sore formed. The improvement that the German navy made in their Enigma machines was something Turing eventually cracked, using the technology of his bombe devices. But at the same time German forces had developed a new cipher, to use in such top communications as High Command messages, and this the simple bombe couldn’t penetrate.

  From this impasse came funding for others at Bletchley to build a far more advanced machine called Colossus, one that was several steps closer to Turing’s 1930s vision. Once it was working, its hundreds of miles of interior cables generated so much heat, apparently, that the Wrens who adjusted its plug settings sometimes had to ask all men to leave the premises, so they could undress enough to work in the heat.

  Turing’s work had been important enough that he was awarded the coveted Order of the British Empire. A memo he wrote to Downing Street about delays in supplies to Bletchley had even resulted in Churchill jolting his principal staff officer with a memo: “ACTION THIS DAY: Make sure they have all they want on extreme priority….” But even so, the Colossus still needed to be laboriously reconfigured almost every day, as the Germans regularly changed the details of how messages were sent. The machine was more than a calculator, but not yet a true programmable computer. It could make only the simplest of sorting decisions on its own. Compared to what Turing had imagined, it was still excruciatingly slow.

  When the war ended, in 1945, Turing was eager to go further. On one of his last days at Bletchley, as he and Joan filed away their hyper-classified work, Turing told her that he’d persuaded the National Physical Laboratory (NPL) outside London to give him a team so that he could, finally, build the infinitely changeable machine he sought. He and Joan had broken up—on calm reflection, she’d decided marriage to a gay man would not be the wisest of choices—but this was a moment to share confidences.

  At first it went well. The head of the NPL was Sir Charles Darwin, grandson of the great biologist, and one of the best-connected science administrators in the country. What Turing didn’t realize at first—though he learned fast—was that although Darwin had been a useful scientist as a young man, and indeed had helped Rutherford with the original solar system model of the atom, he was now a harrumphing, pipe-smoking grandee.

  Darwin seemed to feel that if this odd fellow Turing wanted to build a practical machine, or even a number of practical machines, that was fine. After all, postwar Britain needed such help. But why was he going on about this “universal” machine, and what were all those musings he heard Turing blurt out—it had been difficult keeping him away from journalists—about a computer that could someday write a sonata, or be given mechanical legs and navigate the countryside, or be programmed to learn anything one could imagine?

  They had meetings, and when Turing spoke about the need for vacuum tubes and telephone switching relays, Darwin was interested. When Turing spoke about pure math, he was polite. But when Turing insisted on speaking about software, and suggested that they didn’t need to build a complicated machine, that their goal was to make the machine “do all kinds of different things, simply by programming rather than by the addition of extra apparatus,” Sir Charles decided that this young man did not understand the modern world at all. How could one possibly build a machine by, expressly, not building a particular machine at all? There were rumors that Turing had done something very important in the war, but to Darwin he clearly seemed off the rails.

  Turing was exasperated. The only way to get a computer to operate, he knew, was to create a physical machine that was quite simple, with an arrangement of hardware inside so pliable that one could devote one’s efforts to coming up with ingenious programming that would change how the electrical signals inside were shuttled around. There was no need to build the physical machine afresh each time. His fellow workers had come close to this design with the wartime Colossus, in their efforts to keep up with the constantly changing arrangements inside the German’s coding machines. At Bletchley there hadn’t been enough time to do more. Why wouldn’t anyone let him go further now?

  It didn’t help that Turing still had no sufficiently miniaturized components for his switches or his storage devices. (The software inside his machines could just be a changeable pattern of electric circuits, but there would also have to be solid hardware, such as a memory section to record what the software came up with.) Yet the only components that were available were large and bulky. To help with the memory devices he tried a trick from the radar project of World War II, where technicians had found that if they filled a big round pipe with gallons of dense liquid mercury, and then sent pulses of liquid waves through that mercury, those pulses bounced back and forth inside the pipe with surprising accuracy.

  The pulses seem fast to us, but are so slow on the time scale of electrons that the pattern lasts long enough to form the desired memory device. Yet funding was so poor that at one point Turing was reduced to scavenging the suburban fields outsid
e the NPL labs for fragments of plumber’s pipes to use in his grand computer’s memory. As 1946 turned into 1947, Turing made hardly any progress. Darwin and the other administrators became suspicious. Turing looked for other jobs.

  In the autumn of 1948, Turing moved to Manchester, where word had it that something closer to a true computer was being built. (There were competing projects in the U.K. now, for a number of researchers had picked up on Turing’s prewar papers, or on American wartime work similar to the still-classified Colossus machine. All the teams wanted to take this research further.) But if London and Princeton had been difficult for Turing, Manchester was impossible. The mathematicians designing the device were friendly enough, but they had already started on their own blueprints, which weren’t even as bold as Turing’s original 1937 vision.

  The practical staff in the engineering labs might have been able to modify that design, but they were wary when Turing, with his prep-school, south-of-England accent, introduced himself and tried to get them to help. Everyone at Bletchley had worked together, but that was then. The unity of wartime was fading. The Manchester engineers had a lifetime of experience to show that someone with an accent like his was likely to ignore them smugly, or at least not have any useful advice to offer. Turing knew that he was different. His years with radios and punch-card readers and building electrical systems of almost every sort would have allowed him to give them lots of useful pointers. But with the class system so firmly, invisibly, in place, it was impossible to convince them. Turing was stuck.

  In fact there was a new technology being perfected in America that could have transformed his life—one that involved harnessing the newfound properties of electrons in the submicroscopic quantum realm. Turing had even heard enticing rumors of these devices. But the silent class war meant he couldn’t work with the engineers to chase it down, and anyway, there was no sign of it actually being produced.

  He dawdled for most of 1949, trying one subject after another. At one point he considered trailing back to Cambridge, but becoming an academic seemed petty after all he had done. He tried some of his old topics in pure mathematics, but he was getting too old for that as well. Von Neumann wrote him a cheery note from Princeton (“Dear Alan,…what are the problems on which you are working now, and what is your program for the immediate future?”). But Turing had little to answer. He was alone, perhaps more so than ever before.

  For a while he went back to analyzing the patterns of swirls in daisies and other plants, perhaps remembering those warm days on the lawn with Joan Clarke. He also began to think about what it really meant to be a solitary being—and from that he came up with a paper on artificial intelligence and the nature of self-awareness, which just a few years after his death would be recognized as fundamental to modern cognitive and computer studies. (His ideas on computer simulation of biological development also became important for today’s research.) But in postwar Manchester, where only fragments of the ideal computer he’d conceived were being built, daisies and isolated minds were all just a further sign of this strange man’s irrelevance. Turing was shunted aside once again. Perhaps he accepted it.

  His mother continued to write him regularly, apologetically asking about his progress in finding a wife. It was getting harder to write back with the conventional lies. His love life was meager, and perhaps in memory of the cold showers from prep-school days, Turing had taken up long-distance running, and was at one point ranked as one of the leading marathon runners in Britain. (His best time was within seventeen minutes of the then-current Olympic record. He would have been considered for the British Olympic team but for an unexpected hip injury.)

  Instead of real love he was resigned to occasional casual pickups. They were of no great significance until, one evening in January 1952, Turing realized that a young laborer who had stayed over must have passed information about his house to an accomplice, for when Turing arrived home, he saw he’d been burgled. He went to the police to complain, probably more hurt by the breach of faith than by the cash value of what was taken.

  It was a terrible mistake. Homosexuality was a crime in Britain then. Possibly in Cambridge it might have been handled with a reprimand; in London not long after, the actor John Gielgud would be arrested for a similar offense, yet lobbying by his friends allowed him to minimize time in jail, and newspaper reports were not drawn out. But Manchester was neither London nor Cambridge. The accomplice was arrested, immediately implicated Turing in exchange for immunity, of course, and Turing was seized. He was soon in court, alone, for what was then considered a very serious crime.

  Because of his wartime service—the award from the British government; hints that he might be selected for a knighthood—it was arranged that he could avoid prison. But this meant agreeing to undergo experimental treatment to “cure” his homosexuality. The treatment was mandatory consumption of female hormones. There was little alternative to accepting. Prison would have been brutal, and only at a university could Turing continue his work.

  Turing began the course of pills, picking them up at regular intervals as required. At first he thought the effects would be insignificant, but the treatment made it hard to concentrate. Even if he had tried to get the dosages lowered, the judge was implacable and wouldn’t have allowed it. And then, as the hormone treatment went on, Turing found to his horror that he was developing breasts.

  It was too much. He had no intellectual companions, he had no chance of love—and now his body and mind were being destroyed. The treatment stopped in April 1953, but he never really recovered. One rainy June evening in 1954, Turing was at his home in the Manchester suburb of Wilmslow. He took out an apple, and then he opened a jar of potassium cyanide that he kept for gold-plating reactions in his electrical experiments. The jar still had some cyanide inside when it was examined the next day; the apple, found beside his body, was missing several bites.

  10

  Turing’s Legacy

  NEW JERSEY, 1947

  What Turing was hunting for—which might have saved his life—was right under his nose all the time. He’d even received a hint in 1948, even before he moved up to Manchester. A friend from Bletchley days, Jack Good, had written him:

  “Have you heard of the TRANSISTOR (or Transistor)? It is a small crystal alleged to perform ‘nearly all the functions of a vacuum tube.’ It might easily be the biggest thing since the war. Is England going to get a look-in?”

  But then there was only silence. Something had happened in America to slow down the new device’s development, and the reason goes to the heart of how it worked.

  When Turing was a student, most electricity specialists felt that all the substances in the world were divided into two quite different types. There were substances such as steel or copper that could transmit an electric current, and there were substances such as glass or wood that could never conduct a current. The first group were called conductors, and the second were insulators, They had as much in common as aardvarks and coal mines: the two categories just did not overlap.

  That simple distinction seemed to explain the age-old question of why you can see through glass but not through steel. The inside of a steel wall looks a bit like a huge, abandoned Egyptian temple, full of neatly aligned pillars of iron and carbon atoms. Seen in close-up, however, those atoms aren’t smooth or neat. Many have lost their outermost electrons entirely, and—as we saw in the first radar chapter—those electrons float aimlessly inside the steel. When light flies in, it steadily gets used up as it shoves those electrons so that they start moving with greater energy. This means that as a light wave goes deeper into the steel, there’s less of it that hasn’t been diverted—“soaked up”—by those lurking free electrons. It’s as if a wave of explorers entered the abandoned temple and, one by one, were dragged behind the pillars. Pretty soon there aren’t many explorers left. Light waves reflected from you can go in one side, but won’t make it out the other.

  Inside a wall of glass, by contrast, the atoms are better
behaved. Their outermost electrons are much more tightly bound to their atoms and have no interest in waylaying explorers. A light wave that enters their domain will soar in and out unscathed, emerging as bright as ever. Light waves reflected from you will travel all the way through—you can be seen by someone standing on the other side.

  That difference is why metals conduct electricity, and glass doesn’t; indeed, it is why electric wires often sit on glass insulators. A current hums easily through the copper or aluminum in those wires, because the metals have so many free electrons inside. The pushing force, the invisible whirlwind from the power station, just nabs and pushes them along. The current doesn’t pass across the glass insulators, however, for glass doesn’t have free electrons inside to become a moving current. A live electricity pylon is like a very, very stupid switch: it’s always in the “on” position, for electric current just runs forward, along the wire, and never switches direction to go downward through the glass insulators the wire rests on.

  If there were only these two possibilities—that some materials can always conduct electric current, and others can never do that—then Turing’s legacy would consist only of a few interesting papers and some huge, constantly overheating rooms filled with plugs and vacuum tubes in complex arrangements. The computers we take for granted today wouldn’t exist. But metal and glass aren’t the only substances that exist.