DAYS AFTER BUCK GOT BACK FROM BERLIN, HE PACKED HIS BAGS and moved to Boston. His trip to Germany and Austria under the aegis of the CIA had constituted the last two weeks of his two years’ obligatory service to the US Navy.
There was a growing list of corporations building computers or computer components. Many of them, such as RCA, had graduated from building radio sets and broadcast equipment. Remington Rand, the typewriter manufacturer, was investing in a big way. Burroughs, a pioneer of adding machines, was starting to get involved. Presper Eckert and John Mauchly, the two principal academics who had built the ENIAC, had also set up a business, and dozens of other university-based researchers were looking to do the same thing.
The public face of the new computing industry was Thomas Watson, the president of IBM. He had been the first to fire the public’s imagination about computers when he had unveiled the Harvard Mark I, a joint effort between IBM and Harvard University, to the world’s press in 1944.
Watson requested that the machine be fitted with a stainless steel casing covered in bright, flashing lights that would make it look more futuristic and appealing. It was entirely unnecessary but he hired the designer Norman Bel Geddes to produce the casing, at an additional cost of about fifty thousand dollars.
Science fiction writers were beginning to latch onto the importance of computers, thanks to these high-profile stunts. Yet most of the work in the field was still emanating from the three branches of the military—the army, navy, and recently created air force.
Buck had long had his eye on a place at MIT. For any young scientist trying to carve a path in the new world of computer technology, Jay Forrester’s lab at the prestigious university was considered one of the best places to be. Forrester was assembling a team to build the world’s biggest and fastest computer. He boldly named it Whirlwind in anticipation of the record speeds he expected it would achieve.
Early in 1950 Buck applied for a place at the MIT graduate school to study electrical engineering. He was rejected; MIT complained that his grades had not been high enough during his time in the V-12 program at the University of Washington. A snooty letter from MIT admissions suggested he may be better suited to finding work as a lab assistant, which could perhaps allow him to take a couple of classes; a full-time course would be too demanding for him, it said.
Buck appealed. He had been in the top quartile of his class at the University of Washington, and his class had been formed from the top 10 percent of America’s high-school students. He had the grades.
After a quiet word from Seesaw, and a few high-powered recommendation letters, Buck was pushed through the MIT admission process and installed on Forrester’s flagship Whirlwind project.
Buck started as a research assistant in Forrester’s lab on July 3, 1950. His salary, paid out of the newly secured Whirlwind budget, was $172.50 per month. The tuition for his degree was paid for by the US Navy, and he was still in the navy reserve. The new salary he was receiving was also coming mostly from the navy, indirectly, given that it came from the Whirlwind budget. Thus, Buck was still closely connected to the military.
MIT had a rich heritage of wartime discoveries, most notably in developing radar technology. It was still involved in dozens of military projects, of which Whirlwind was one of the most costly. Forrester had been influential in securing the university’s position in these projects, having been one of the first American academics to spark a fire of enthusiasm for computing technology at the highest levels of government. He had helped to show that computers could be much more than just adding machines that could spare the pencil scribblings of mathematicians.
In 1948, just before Buck started at Seesaw, Forrester published a seminal paper that raised the bar substantially in terms of thinking about what these new machines could be used for.
Forrester and colleagues’ Forecast for Military Systems Using Electronic Digital Computers provided a series of fifteen-year forecasts on how technology could evolve—much of which sounded like fantasy at the time. The report predicted that computers would soon be able to guide missiles to specific targets, and spoke of computerized planes, machines that could transmit data around the world at high speed, and computers that could intercept Soviet atomic bombs. It also suggested that computers could revolutionize industrial processes with the use of robots. It was, in other words, a remarkably prescient piece of analysis.
To work toward these transformational goals, Forrester said, the US military would need to spend a staggering $2 billion between 1948 and 1963—about $20 billion in today’s money. Forrester’s proposal was labeled interesting, but expensive, and was thus pushed aside for two years. By the time Buck joined the team, however, the need to develop bigger and faster computers had become more pressing.
Whirlwind had started its life during World War II, under the code name Project Kiddy Car. It was commissioned by the US Navy, and was originally intended to be a state-of-the-art flight simulator for training wartime pilots (the US Air Force only came into being after the war, with planes the sole concern of the navy until that point).
Before it was even built, interest waned in the flight simulator idea—largely because by then the war was over, and the need to produce a high number of pilots was substantially diminished. Forrester was reluctant to let go of his funding, however, so the design was repurposed: the as-yet-unbuilt Kiddy Car computer would be developed as a computerized air traffic control system instead, the first of its kind. While those may sound like very different concepts, at their core both systems were about computing rapidly changing data about the speed and trajectory of objects flying through the sky.
The project was daring. Up to that point, computers had been devices that could be set up to run large calculations and then churn out an answer. Data would be fed in—whether it be a German encryption code, or the trajectory of an artillery shell—and a room full of men in white lab coats would wait for the answer to pop out, typically on a piece of printed tape or maybe on the screen of an oscilloscope. They often had to wait for hours. The computer would quite often break down halfway through because a valve blew up.
Project Whirlwind was different. Whether it was a flight simulator or an air traffic control system, it had to perform its calculations in real time and react to new information as it became available. To say that the technology did not exist to do the job is beyond understatement. Yet that was the whole point.
Forrester’s team had been trying all kinds of new ideas to build this fantastical new machine. They were struggling to find components that could process data quickly enough to do the job. Their experiments were also chewing up a huge amount of money—about $600,000 a year.
By the spring of 1950, just before Buck’s arrival, the navy was beginning to lose patience. Yet thanks to the rising threat of communism, there was a new sponsor for Forrester’s cause—the Air Force.
Now that the USSR had its own atomic bomb, there was an obvious problem to address: how could a Soviet bomber be stopped from encroaching on American air space and delivering a deadly payload? George Valley, an MIT physics professor who had worked in the university’s famous radiation lab during the war, had been commissioned by the Air Force to pool together a team of technical experts who could build some kind of radar machine to anticipate and block the nuclear threat. Over lunch on the MIT campus, Forrester convinced Valley that the Whirlwind system he was building could be capable of reading signals from the Doppler radar stations around the United States, serving as an early warning system for a Soviet attack.
On the back of Valley’s recommendation, the air force started contributing to Whirlwind in April 1950. It was willing to put up $200,000 of the $600,000 annual bill. The reduction in spending was enough to keep the navy happy, and thus the scheme became a joint effort between the air force and the navy.
The mission was to build a computer that could track enemy aircraft, distinguish them from US air force planes, and calculate their speed and direction. Once the course of an invading bomber had been identified, exact coordinates would need to be fed to a US fighter plane that could be dispatched to intercept.
Forrester had clearly promised more than he could be sure to deliver. Yet his timing was better than even he could have realized. A few months after securing support from the Air Force, America was at war again, this time in Korea. Communist-controlled North Korea had invaded South Korea.
No one in the US intelligence agencies considered Korea a likely flashpoint. There was just one American soldier stationed at the border when the North Korean tanks started rolling in his direction. The failure was colossal. America’s spy community was ripped to shreds by a series of committees, inquiries, and investigations. US intelligence had been building up its networks in Europe to deal with the Soviet threat, but now America was engaged in a land war with the communist bloc in a country many American voters had never even heard of. While intelligence officers had been training students in German and Russian, there was now a need to find Chinese and Korean speakers—and quickly.
Paranoia ensued. Seesaw was cleared to take its staffing levels back up to wartime levels; it was given new responsibilities too. As a result of the intelligence failure, codebreaking and signals intelligence was centralized in an attempt to end the infighting among different agencies. It was Seesaw that won. The organization went through several rebranding exercises over the next few years, but Buck’s former commanding officers remained in prominent roles all the way through.
The immediacy of a new war ensured there was a favorable wind behind any new technical tools that could give the United States the edge in the battle against communism. Given its clearly defined purpose as an anti-Soviet defense tool, Whirlwind became of elevated importance—along with countless other computer projects across the United States. All branches of the military could see that computers were part of a future centered around nuclear bombs and long-range attacks. And the handful of individuals who knew about the technology were of prime strategic and military importance.
Buck arrived at MIT on July 3, 1950, just eight days after the North Korean tanks rolled across the thirty-eighth parallel. Thus, it was in this period of additional enthusiasm for the project that he found his feet.
His first job at Whirlwind was on its input and output systems: the ways and means of inputting data to the machine and having the results read back to the user. On a modern tablet computer or smartphone, both functions can be performed on a touch screen. In 1950, however, the method was far from settled.
Some of the more advanced computers were starting to use electric typewriters to punch numbers into the machine. Most test results were generated through some form of printout. None of that was good enough for Whirlwind—not if it was going to meet its task of identifying and intercepting Soviet bombers bringing deadly payloads to American shores at speeds of hundreds of miles per hour.
For Whirlwind to function, it needed a totally new way of operating. Buck designed a display based on a standard cathode ray tube, like those used in televisions. It was accompanied by a series of modified oscilloscopes, little different from a standard radar display. Buck built in a camera that photographed the screen periodically to acquire accurate information on the trajectory, speed, and bearing of each plane.
He then devised a system to record the radar signals being received by Whirlwind and the voice of the operator sending instructions to the pilot. It allowed the whole sequence to be captured on magnetic tape, and replayed, with the commentary that accompanied the action, as if in a movie. Those sequences could be replayed time and again without exhuasting the computer’s valuable brainpower, as it was all recorded on tape. By the standards of 1950, this was revolutionary. It was swiftly tested, modified, and perfected to the point at which the military felt comfortable placing it into active service.
Yet Whirlwind had to do more than just record signals. The air force officers at the controls of the machine had to be able to pick out individual planes from the soup of information being received, and then calculate each plane’s speed and coordinates. On the Whirlwind computer’s screen, each plane, friend or foe, was represented by a small dot of light. It was up to air force staff to decide which dots were more interesting than others. A speedy method was needed for the Whirlwind operators to pick out the right stream of radar signals and divert Whirlwind’s computing power in the right direction.
Buck sketched out a plan for something he called a target acquisition joystick. Although joysticks had been used to control airplanes since the early days of flight, no one had used one to operate a computer before. Based on the labored explanation that Buck offered for his invention, it seems that it was difficult for others to comprehend the concept, which he explained to his former military commanders in July 1950:
A manually positioned spot on the display oscilloscope was decided upon as a convenient method of telling the computer which of several targets it is to work on. When the spot has been positioned so as to coincide with desired target on the scope face, a signal will be sent to the computer.
It has been decided to experiment with a joystick type of device for manually positioning the spot with a pushbutton in the top of the joystick to inform the computer that the operator is on target.… Motion towards the north, south, east or west or any combination of these is done by pushing the joystick in the desired direction.
It is unclear from the various accounts of the Whirlwind project whether the joystick ever came into service. Buck complained in his notes of encountering problems with the microswitches inside his gadget. Yet he soon moved on to a different invention that solved the problem; it was a computer tool that would eventually be found in many homes in America, Europe, and Japan by the mid-1980s, and on the TV game show Jeopardy: the light gun.
The Buck light gun was a small pen-like device, with a button for a trigger, hooked up to the computer. The Whirlwind operators would just point on the screen at the plane they wanted to track, then click the button. The electron gun inside the tube could then detect the spot identified and keep track of it using Whirlwind’s computing power.
It was broadly the same technology that would later be used for the gray and orange zapper device sold with the Nintendo game console. Whereas the later gadget would be used by ten-year-olds to shoot digital ducks, Buck’s original machine was for targeting Soviet bombers.
IN SPITE OF the enormous challenges at the outset, Whirlwind became an overwhelming success and laid the ground for a system the US military would dub SAGE (for Semi-Automatic Ground Environment). SAGE led to dozens of colossal machines installed in huge concrete cubes at twenty-four different air bases and other military installations. Each computer took up about an acre of land. It was the immediate predecessor to the North American Air Defense Command (NORAD) missile defense system still in use today.
Within a few months of Buck’s arrival at MIT, Whirlwind was starting to prove itself an impressive device. It could display friendly fighter planes with a letter F, and target planes with a T. The point at which they would intercept was shown as X.
The military’s interest in the project remained intense. While the machine was still being developed with a view to air defense, Whirlwind could be utilized for a great many tasks. That in itself made it unusual: at the time, computers were still mostly considered as machines built to perform specific tasks.
For five hours a day, MIT staff were allowed to run their own programs through Whirlwind—to play with their toy and find out what else it could do. It could run about one hundred programs in that time. While it was being operated by the military, progress continued apace.
The defense chiefs seem to have been keen to brag about the creation that their endless stream of checks had spawned. Whirlwind was more impressive than anything the Russians had built. In what appears to have been a textbook lesson in Cold War propaganda, the Whirlwind machine came to be demonstrated live on prime-time American television.
On December 16, 1951, the Whirlwind machine—and Forrester, its handler—were interviewed on the CBS flagship news show See It Now, the main vehicle at the time for Edward Murrow, one of America’s most famous newsmen. He shot to fame in World War II with his rooftop broadcasts from London during the blitz, and his coverage of the liberation of the Buchenwald concentration camp in Germany. Later he would go on to lead a campaign against Senator Joseph McCarthy’s communist witch hunts—an episode of his career that was turned into an Oscar-nominated 2005 film directed by George Clooney, Good Night, and Good Luck, named after Murrow’s wartime catchphrase.
The broadcast had the tone of a propaganda reel. Even the advertisements selected to run alongside the segment appeared to have strategic significance; immediately before Forrester’s brief moment of televisual fame there was an advertisement for aluminum manufacturer Alcoa, boasting about how it was now producing four times as much aluminum as it had in 1939.
“These are days of mechanical and electronic marvels,” said Murrow as he teed up his interview. “The Massachusetts Institute of Technology has developed a new one for the navy. It’s the Whirlwind Electronic Computer.”
Those last three words were uttered slowly and carefully: the term computer was still about as familiar to CBS viewers as the more obscure entries in a medical dictionary.
“With considerable trepidation we will now attempt to interview this machine,” Murrow continued. A live video feed appeared on screen, showing the oscilloscope display of Whirlwind in the lab at MIT. It was flashing the words “Hello Mr. Murrow” with its tiny lights.
“I assume like any other piece of delicate electronic equipment, there is a human element to this,” continued Murrow.
“Yes, Mr. Murrow,” said Forrester, who came into shot as the camera panned out. A light suit hung from his angular frame. He was perched awkwardly on a stool, surrounded by banks of dials, gauges, and wires. He kept one hand clutched to the earpiece feeding him instructions.
Forrester proceeded to give a tour of the machine, making particular effort to point out its storage tubes, which could process data in twenty-five millionths of a second. As the camera panned across the lab, the massed ranks of electrical circuits that made the machine work were displayed.
Forrester then asked if Murrow would like to ask the machine a question. Since the navy had paid for the machine, Murrow said, wouldn’t it make more sense to ask them to set a task? On a second monitor behind Murrow’s desk, the navy’s head of research, Admiral Bolster, had been patched in on a second live video feed.
Bolster proceeded to outline a mathematical puzzle to calculate the speed, trajectory, and fuel consumption of one of the navy’s Viking rockets. It was all part of the show; the calculation had been preprogrammed into Whirlwind. It showed one bar of lights on the left-hand side of the screen, representing the remaining fuel in the rocket, and another on the right showing its speed, with the flight path charted in between. Forrester and the admiral looked incredibly smug about their work.
Murrow then asked his own question: if he had been the “Indian”—that is, the Native American—who sold Manhattan island for twenty-four dollars in 1626, how much would he have now, assuming he invested the twenty-four dollars and had received a constant rate of return of 6 percent? Demonstrating the machine’s versatility, Forrester produced a preprogrammed paper tape with punched holes containing the problem. He fed it into a reading device on a giant control panel, then watched while an electric typewriter on the other side of the room typed out the answer. Murrow was clearly impressed. Anticipating the advent of accounting software, the reporting legend then asked if the machine could also work out his tax bill.
Forrester appears to have been thrown by this minor improvisation, which deviated from the script. For his parting shot, however, the lab boss produced “another type of mathematical problem that the boys here have worked out in their spare time.” With that, Whirlwind started to play a tinny, electronic version of “Jingle Bells”—bringing a smile to Murrow’s face.
All of these computations were straightforward for Whirlwind. The broadcast made no mention of the computer’s real work for the air force, or the complex early warning system it was running. Yet the demonstration could be assumed to have served its purpose, in terms of proving technical superiority to the Soviets. The television demonstration had proven that Whirlwind was a multipurpose machine.
Its main achievement had not been visible from any of the party tricks performed for CBS viewers. It was the memory system that was driving the machine. Within a few short years, the titans of corporate America were not just copying the Whirlwind’s revolutionary technology but claiming that they had invented it first. It was thanks to his exposure on this project that Buck became obsessed with trying to build the fastest, smallest computer possible. He would soon start work on his own solution to the memory problem that would see his name suddenly soar in prominence within the broader scientific community.