JACKIE BUCK WAS IN SHOCK. THE DEATH OF HER HUSBAND, Dudley, was just too much to take in. Her mother had come to look after the kids, who were far too young to understand what had happened. David, who was just eight weeks old, would never know his father. Doug’s only memory of his father is that of the funny truck with the red light on top that had taken him away.
A police cordon had been set up around the house. The NSA was a little uncomfortable about what might be in Buck’s basement office. Everything he was working on was sensitive, but it was possible Buck had taken home work related to the Corona spy satellites, or details of his talks with Lockheed on missile guidance systems. There would certainly be papers linked to Project Lightning, the supercomputer project based on Buck’s cryotron.
MIT was also uneasy; documents crucial to future valuable patents could well be stored in Buck’s basement. Gordon Brown and Ewan Fletcher, Buck’s two superiors from the electrical engineering department, were dispatched to Wilmington, Massachusetts, to remove everything even vaguely work-related from Buck’s house on Birchwood Road. It was only one day after his death.
“They came and they just took all his notebooks,” remembers Jackie. “I didn’t have a choice. They just came into my living room, and sat down. I wasn’t sure why they were there, and they said they needed all of the notebooks back and they intended to take them that day. Dudley also kept a folder full of ideas that he wanted to explore. When I went to his files a few days later, they had taken that too—or someone had taken it.”
In Building 10, the record player was silent. It was Carol Schupbach, Buck’s secretary, who had broken the news. After they hadn’t heard anything about Buck for a couple of days, she had called the house. Jackie’s mother had answered the phone. Carol’s piercing scream was enough to let everyone know what had happened.
“We knew Dudley was sick, we heard he had been taken to hospital the day before,” remembers Bernie Widrow, who worked in the office neighboring Buck’s in the lab. “Carol Schupbach let out a scream like, ‘Oh, my God almighty,’ and I figured it out in an instant. Nobody expected that. Two days before he was in the lab working away. He wasn’t feeling well. Everybody was telling him, ‘For Christ’s sake, why don’t you go home, get some rest?’ The next day he didn’t come in. The next day he was taken to hospital, and the day after that he died.”
The group of young students that had gathered around Buck was particularly hard-hit. “It was one of the saddest moments of my life, because to me he was my hero, he was a man to look up to, a man that almost substituted for my own father,” remembers Allan Pacela.
It fell to a heartbroken Chuck Crawford to call Ken Shoulders out in Stanford. “I was shattered, everyone was shattered. I called Ken, and he was just speechless.”
After a few days of shocked disbelief Shoulders, in particular, started to ask questions. The cause of death was officially recorded as beta-hemolytic streptococcus, a throat infection that can cause tonsillitis, rheumatic fever, and scarlet fever, and in rare cases evolves into a flesh-eating virus. The theory was that it had been lurking in his body for weeks, but that Buck had been fighting it off quietly. Yet there was never an autopsy conducted to prove or disprove the theory; at the time, there was no legal requirement to conduct a postmortem examination, and Jackie couldn’t bear the thought of Buck being cut up. “I didn’t do an autopsy,” she recalls, adding,
He was so vibrant, so young, and I was so shocked. Looking back, he should have had a proper autopsy. I just didn’t want to think about that. I was naive; I was only twenty-five when he died.
The truth is that he died so quickly after he was in the hospital that they didn’t really have time to get the lab results and the blood work. That’s why there was such a mystery as to why he had died. MIT was concerned he had died from something pneumonia-like in his lungs. It could have been something there that could have given him the semblance of something pneumonic. They were concerned they had not had something hooded properly, or something like that.
MIT’s official recorded version of Buck’s death, uncovered during the research for this book, is extremely strange. A letter sent by Hardy, the university doctor, to Ewan Fletcher claimed that Buck had become ill sometime on the weekend of May 9 to May 11 and had been off work most of that week.
Dudley and Jackie had looked after the three children of their neighbors Nancy and Paul Bodenstein that weekend, in addition to their own three. Buck had been absolutely fine. Contrary to the letter, he had not taken a single day off work until the day his package arrived. While some former students interviewed in the course of our research have claimed that he had looked ill during lectures, and that he was lacking his usual enthusiasm, that does not tally with the recollections of those who worked closest to him. Everyone from his lab said it was only late that Monday afternoon, after the package arrived, that Buck had complained of feeling ill. More important, Jackie had seen nothing wrong with him.
As the father of a newborn, working all hours of the day to finalize a groundbreaking invention while simultaneously handling dozens of classified government projects, it is understandable that Buck may have looked a little tired. But to his nearest and dearest he had not complained about anything.
An MIT faculty meeting on June 9 resolved to “place on its records its own deep sense of loss” and to pass a copy of its memorial memorandum to Jackie. It referred to his “sudden and untimely death,” almost contradicting the official medical account prepared by the university’s own doctor, and then went on to praise his work:
During his nine years at the Institute, Dr. Buck made outstanding contributions to the field of electrical engineering, chiefly in applications to computers of solids and their low-temperature properties.
In 1957 the Institute of Radio Engineers awarded Buck the Browder J. Thompson memorial prize for his paper on the development of the cryotron, the first miniature electronic device making use of superconductivity at extremely low temperatures. This breakthrough in low temperature superconductivity has been of great significance in the world of electronics in general and to computers in particular. During the past two years Dr. Buck had carried miniaturization even further, and at the time of his death was seeking to make cross-film cryotrons with dimensions of only a few millionths of an inch.
Buck’s loyalty to the department and the “thoroughness” of his teaching were praised, as was his work on recruiting trips to high schools, noting that “during these tours he demonstrated what one observer termed a ‘contagious quality of optimism, enthusiasm and just plain joy about each man’s particular work.’” The memorandum continued, “Dudley Buck’s originality and creativity were a constant source of wonder and admiration to his students and to his associates. He combined a warm heart with a lively intelligence of great reach, and a seriousness of purpose with the enthusiastic energy necessary for its implementation. He had all the attributes of greatness and before his untimely death had already achieved much. His loss will be keenly felt by his profession and by his colleagues at the Institute who loved and admired him.”
Ken Shoulders was never satisfied with the official explanation of Buck’s death. He was convinced that Buck had been poisoned by something in one of the chemicals that was used in his experiments—perhaps some of the gasses that were released by the chemical reactions. He eventually zoned in on tetraethyl orthosilicate, believing that exposure to the compound had created a condition in his lungs that looked like some form of pneumonia.
David Brock at the Computer History Museum in Mountain View, California, believes Buck’s death could have been caused by exposure to poisonous gasses. On that final day in the lab he used boron trichloride gas. The process for depositing that gas to make a test microchip releases hydrogen chloride. Exposure to either gas could cause fatal pulmonary edema, leading to symptoms similar to pneumonia.
Yet Chuck Crawford disagreed with Shoulders’s view at the time. Since then he has disagreed with anyone who has posited similar suggestions of chemical poisoning.
If Buck had been killed by something from the experiments, then surely Crawford would have contracted it too. He was the one doing much of the hands-on work in the lab, and he had been fully checked out after Buck’s death.
Crawford is no conspiracy theorist. He had no idea about Buck’s second life with US intelligence and security services until being contacted in regard to this book. There was a long incredulous pause on the end of the phone as Buck’s dealings with the NSA, his encounters with Soviet scientists, and his work on top-secret projects was explained.
For Crawford, Buck’s frequent trips away had always been explained away as MIT business or science conferences, if they were explained at all. It’s certainly true that the government work overlapped with his science, in any case.
The lab team in Building 10 did not know that, four days after his death, Buck was due to fly to Washington for the first meeting of Ridenour’s new computer advisory group.
Is it mere coincidence that Buck died twenty-nine days after Khrushchev’s top computer scientist, and six of his colleagues, were given a guided tour of the lab? That might be credible were it not for the fact that Buck was not the only elite America computer scientist to die suddenly and mysteriously that day.
LOUIS RIDENOUR CHECKED in to the Statler Hotel in Washington, three blocks north of the White House, on May 20, 1959. He had flown in from Los Angeles a few days ahead of his key summit at NSA headquarters, presumably to squeeze in some Lockheed business.
Two months earlier Ridenour had been promoted to the position of vice president of Lockheed; the company’s new Electronics and Avionics Division had been separated out of the Lockheed Missile Systems business and handed to Ridenour.
A lot of the big government contracts worth winning over the subsequent few years would be directly related to the space race. While there would be rockets to build and satellites to launch, there would also be lots of money to be made from the instruments and control equipment that would be needed. Lockheed, therefore, set up a stand-alone division to handle this work. Having been one of Lockheed’s top scientists for four years, Ridenour had wanted to prove that he could also be an executive “capable of making a profit.” He was certainly close enough to government thinking on computers to know where the opportunities might lie. The work of the new division would be directly tied to his work with the NSA on supercomputers and to Project Lightning.
The morning after Ridenour checked in to the Statler he was found unconscious by a maid. He had suffered a brain hemorrhage at some point overnight. Ridenour died at almost exactly the same time as Buck. He was forty-seven, and in apparently good health.
Time magazine’s June 1, 1959, issue carried obituaries of the two men on the same page:
Died: Dudley Allen Buck, 32, exuberant M.I.T. electrical engineer and miniaturization expert, who developed the tiny cryotron to replace the transistor, was working on a cross-film cryotron (diameter: four-millionths of an inch) that would reduce a computer from room to matchbox size; of virus pneumonia; in Winchester, Mass.
Died: Louis N. Ridenour Jr., 47, top-notch nuclear physicist who, despite being emotional about his specialty (in 1946 he wrote a grim, prophetic, one-act play about flocks of satellite bombs orbiting 800 miles above the doomed earth), pioneered in missile programs as chief scientist (1950–51) of the Air Force, helped develop the Polaris and X-17 missiles as research director of Lockheed Aircraft Corp.’s missile-systems division, became a Lockheed vice president last March; of a brain hemorrhage; in Washington.
Nobody at the time seems to have connected the two deaths—not publicly, in any case.
The links between Buck and Ridenour were through their cloak-and-dagger dealings with the NSA. There would be no reason to assume a connection otherwise: one died from a brain hemorrhage, one from a pulmonary condition; one was in Washington, the other in Boston; one died overnight, the other over a period of days.
There have been some odd references over the years that seem to tie the deaths of the two men together, however. Howard Aiken, the American physicist who worked on the Harvard Mark I computer, claimed to have been with Ridenour on the night he died.
In a heavily redacted oral history interview for the Smithsonian Institution conducted in February 1973, shortly before his own death, Aiken also referred to a mysterious young man that appears to be Dudley Buck:
I was very, very fond of Louis. I was with him the evening he died, in Washington. It was a very unusual thing. There was a young man who was interested in very low temperature devices in computers—what was his name? Well at any rate, he and Ridenour and I talked [redacted], and we [Aiken and Ridenour] went to Washington.
I met Ridenour and we bummed around Washington all evening with him. And the last thing he said to me before he left was, “You’re going to the Cosmos Club. I wish I was going to the Cosmos Club, but I can’t go because I’m now Vice-President of Lockheed and I’ve got that two-room suite,” and he shoved off and [redacted]. And the next morning, I got up at a quarter to eight and Ridenour didn’t show up and this other man didn’t show up. Somebody wanted to know where they were and I said, “Well don’t worry about Louis, he probably has a hangover this morning.” And a couple hours later, the manager of the Statler called up looking for me and telling me that Louis had died in bed. And almost immediately after that, we got a telegram that [redacted] was dead, so we just folded the meeting and everybody went home. It didn’t seem very worthwhile to go on.
It seems difficult to believe that the young man who was an expert in “very low temperature devices” could have been anyone other than Buck. He was the inventor of low-temperature devices, and was on the panel for that reason. Why his name has been redacted from this record is a curious mystery.
Sergei Khrushchev refuses to believe that his father sanctioned any kill orders on American scientists. “Assassinations and all these things were not part of the behavior at this time,” he explains. “Now you can assassinate anyone, anywhere, using drones and other things. At that time, you needed exceptional things. To assassinate one of the scientists in the United States, it is such a crazy idea I don’t think anyone could discuss it—not only in government, but in the KGB itself.”
He is even more adamant that Lebedev, Glushkov, “or any of these individuals” who came to visit Buck would never have been complicit in any such plot. Khrushchev posits that he should know: he had encounters with all of these men while working on the Soviet missile program.
“They are not going to say, ‘These are my competitors, please kill them,’” he adds. “When I was working with Russia’s missiles, we tried to do our best. The Americans tried to do their best. It was beyond our imagination to think about this type of thing. Forget about this conspiracy theory. I, one hundred percent, do not believe in these things.”
It is certainly true that Soviet executions of foreign agents happened more frequently in Ian Fleming novels than in real life. Nonetheless, they did occur. That notion has fed into the greatest conspiracy theory of all time.
A CIA paper on Soviet killing and abduction techniques was produced in February 1964 for the President’s Commission on the Assassination of President Kennedy. The document, which runs through the repertoire of KGB assassins of that time, was declassified in 1993. It provides evidence that Soviet agents knew countless ways to dispatch targets without leaving any detectable sign of an assassination: “It has long been known that the Soviet state security service (currently the KGB) resorts to abduction and murder to combat what are considered to be actual or potential threats to the Soviet regime. These techniques, frequently designated as ‘executive action’ and known within the KGB as ‘liquid affairs’ can be and are employed abroad as well as within the borders of the USSR. They have been used against Soviet citizens, Soviet émigrés, and even foreign nationals.”
The CIA knew of several assassination operations since Nikita Khrushchev’s rise to power, confirming that “the present leadership of the USSR still employs this method of dealing with its enemies.” The tendency to bump off rivals had not died with Joseph Stalin. The report further notes,
The sudden disappearance or unexpected death of a person known to possess anti-Soviet convictions immediately raises the suspicion of Soviet involvement. Because it is often impossible to prove who is responsible for such incidents, Soviet intelligence is frequently blamed and is undoubtedly credited with successes it actually has not achieved. On the other hand, even in cases where the Soviet hand is obvious, investigation often produces only fragmentary information, due to the KGB ability to camouflage its trail. In addition, Soviet intelligence is doubtless involved in incidents that never become officially recognized as executive action, such as assassinations which are recorded as accidents, suicides or natural deaths.
Based on confessions from former KGB agents, the paper claimed that the Russians had been using highly advanced, untraceable poisons that could simulate death from assorted medical conditions since at least 1957, two years before Buck’s death.
Bogdan Stashinsky, a former KGB assassin, admitted that he had killed the Ukrainian writer Lev Rebet in Munich with “a poison vapor gun which left the victim dead of an apparent heart attack,” the CIA document shows.
He delivered the poison with a small aluminum cylinder approximately fifteen centimeters long and three centimeters in diameter that weighed about two hundred grams. It had a fine metallic screen over one end, and contained a liquid poison, hermetically sealed in a small plastic container. A small explosive charge in the other end of the cylinder could be activated by a detachable trigger that then drove a piston inside the cylinder that smashed through the capsule. As soon as the poison was exposed to the air, it vaporized. Agents were advised to use the tube at close quarters, just a few inches from the subject’s face. But it was effective from a range of up to about fifty centimeters.
As the CIA report notes,
The effect of the poison vapors is such that the arteries which feed blood to the brain become paralyzed almost immediately. Absence of blood in the brain precipitates a normal paralysis of the brain or a heart attack, as a result of which the victim dies. The victim is clinically dead within one and a half minutes after inhaling these poisonous vapors. After about five minutes, the effect of the poison wears off entirely, permitting the arteries to return to their normal condition, leaving no trace of the killing agent which precipitated the paralysis or the heart attack.
Allegedly, no foreign matter can be discovered in the body or on the clothes of the victim, no matter how thorough an autopsy or examination. The liquid spray can be seen as it leaves the weapon, however, and droplets can also be seen on the face of the victim.
Stashinsky confessed to using the same type of weapon against exiled Ukrainian nationalist politician Stepan Bandera in October 1959—a death that was also officially recorded as a heart attack. Bandera was part of Reinhard Gehlen’s secret spy network, the same CIA organization that Buck had been assigned to on his two trips to Germany.
The Russians had also developed a weapon that they called a “noiseless gas pistol,” the Kennedy assassination inquiry papers claim. Powered by a three-hundred-volt battery, it fired a lethal, odorless, unidentified gas that was effective from a distance of up to twenty meters and took between two and three seconds to act. It was a contact poison that was absorbed through the skin and was still effective through layers of clothing.
In March 1955 the Russians used poison in an attempted abduction of Lisa Stein, an interviewer for an American propaganda radio station in West Berlin. She was fed candy doped with a poisonous drug called scopolamine. She was expected to fall ill on her way home from the café where she had met her contact. She would then have been picked up by a waiting car. She did not become ill until she was almost in her apartment, however, where neighbors were able to get her to a hospital. After forty-eight hours of severe illness, she was fed an antidote.
One other example of a poison method appears to have been included in the memorandum, although this was never declassified. Yet the CIA recognized that they did not know all of the Soviet techniques: “There appears to be no consistency in the use of poisons by Soviet intelligence to cause disability or death, or in the repetitious use of any one drug. Chemicals which have been used in cases known or suspected to be Soviet-instigated include arsenic, potassium cyanide, scopolamine, and thallium. Other likely substances are atropine, barbiturates, chloral hydrate, paraldehyde and Warfarin. Combinations of two or more substances may also be used, which further complicates diagnosis and tracing.”
The KGB clearly had highly trained chemists on its books, devising ever more advanced poisons. Based on the evidence, it is certainly credible that they would have been able to devise a deadly agent that could be concealed in Buck’s shipment of chemicals. Although Chuck Crawford was with him as they sorted through the chemicals, it was Buck who opened each of the assorted bottles and vials as soon as the box arrived.
It’s equally possible that one of these advanced weapons was used in a more conventional manner. Buck lived a normal suburban life and worked on a big university campus full of people from all over the world. He was not penned-in behind security guards and high fences. If the KGB had wanted to get him, it could have. After all, agents from Amtorg, the Soviet export agency, had found Buck’s lab seven years earlier.
It seems unlikely that the group of Russians who came to visit Buck were directly involved in his death. Yet they were clearly very senior figures who had the ear of the Kremlin.
Former KGB agents, such as Oleg Kalugin, the former bureau chief in Washington, have claimed that all Russians who came to the United States—even students—were actually spies. Of all the things the Soviet delegation heard about on its trip to the United States, Buck’s inventions would have been by far the most surprising.
Soviet intelligence would already have had Buck marked as a potential inventor of the nuclear missile guidance system thanks to his appearance in Life magazine. Given that double agents were operating on both sides of the iron curtain, there is every chance that the KGB had been leaked a copy of the list of attendees for President Dwight D. Eisenhower’s new committee on supercomputers.
At the time of Buck’s death, the KGB was under the leadership of Alexander Shelepin. Countless histories of Khrushchev’s reign portray Shelepin as operating somewhat outside the control of the Soviet leader. Khrushchev biographer William Taubman, for example, has described how the KGB chief ran dirty tricks campaigns against the likes of CIA chief Allen Dulles without the knowledge or consent of the Kremlin. Shelepin was later a key figure in the 1964 coup to depose Khrushchev.
The CIA paper concluded that by 1964, the time of its writing, the KGB would only resort to murder “in the case of persons considered especially dangerous to the regime.” As the designer of the guidance system for America’s intercontinental ballistic missile, and the architect of its most promising new computer devices, who had also been intimately involved in America’s efforts to put spy satellites into orbit, Dudley Buck was arguably more dangerous to the Soviets than anyone else.
CHUCK CRAWFORD WAS supposed to spend the summer of 1959 at the Los Alamos National Laboratory in New Mexico. Buck, ever keen to see his protégés spread their wings, had given him a glowing reference months earlier to help him secure the position. He stayed until the end of July, then MIT called and begged him to come back; after Buck’s death there was no one at the university other than young Crawford who knew enough about the cryotron program.
The whole research program was severely hobbled by Buck’s death. The students were passed on to other thesis advisers and allowed to continue their work, but without Buck to lead the way, the program started to drift. Those who had been closest to the charismatic young professor were given some leeway around deadlines and exams.
“After Dudley died, I continued in the lab, and the amazing thing was my job didn’t go away—for reasons that I never understood,” Pacela recalls. “He was gone, and I cried, I missed him, and he went away, but the job was still there and I could work in the lab, and turn in my hours, and I began to work on my own projects.”
The NSA ensured that research into the cryotron continued. In 1960 Horace Mann at TRW, an electronics firm that had been involved in the construction of the Atlas and Titan missiles and went on to play lead roles in the space race, filed a patent for manufacturing cryotrons using Buck’s method. He filed a related patent later that year, which was assigned to Space Technology Laboratories, the company working on the scientific payload for NASA’s Pioneer spacecraft.
Mann then modified the original patent again in 1961, explaining in more detail how to make cryotron microchips.
After Buck’s death, work also continued at IBM, where there was still a team of at least a hundred set up to build cryotron computers as part of Project Lightning. The team was also working with electron guns and thin layers of chemicals to build ever faster cryotrons; by 1961, two years after Buck’s death, it had built a functioning fortybit memory chip from 135 cryotrons. There were still issues with the production, however. A group was set up that changed the design, but it somehow killed the speed, with this new cryotron switching between one and zero at speeds a hundred times slower than expected.
Between 1961 and 1963, IBM worked on a project for the US Air Force to build a cryotron memory that could be used to derive associations among different types of data. That, too, ended up being scrapped.
By this time both Jack Kilby and Robert Noyce had gotten their semiconductors working. The silicon age of computing had been born. The cryotron, with its need for helium tanks and superconducting temperatures, was overtaken.
Many of the techniques used to make these new integrated circuits were the same or extremely similar to those deployed by Buck. In the case of some of the techniques, it looks like Buck could have gotten there first. The key element in Intel cofounder Noyce’s integrated circuit patent was the use of an “insulating oxide.” It appears that Buck had come up with this idea—and lectured about it extensively—before Noyce got to it.
Buck, however, never got around to filing his patent. Had he done so, a lot of the value in Noyce’s multibillion-dollar patent would have been undermined. The earliest integrated circuit patents from Kilby, meanwhile, make no mention of this key part of the technology.
It took years for Kilby and Noyce to be credited as the joint creators of the microchip. Kilby eventually won the Nobel Prize for Physics in 2000 on the back of the invention. Noyce was dead by this time, and the Nobel Committee does not make awards posthumously. Many more have a legitimate claim for a share of the glory. Had Buck lived just a few months longer, he would have had the opportunity to advance his work to the point where his contribution would be beyond debate.
MIT knew that Buck could have shared in the commercial spoils of this discovery. From the moment the university seized his notebooks from the basement office in his home in Wilmington, it worked through everything it could find to see if there was enough to justify a patent. Eventually the case was closed, with a memo from the investigating scientists to Ewan Fletcher in November 1960, some eighteen months after Buck’s death:
I have gone over this material with Ed Thomas of the Lincoln office, who evaluated some of the Buck and Shoulders work in this area a year or two ago. The conclusions he reached at the time still appear to be valid. Namely: (1) That the scientific phenomena were known at the time Buck commenced work; (2) That Buck had laid out a promising area of experimentation; (3) That the actual inventions upon which valuable patent coverage might be obtained would lie in the work still to be done.
While continued work by Dudley Buck might well have led to valuable inventions in this area, I am sorry to have to say that it appears very doubtful that any valuable protection could be based on the work completed before his death.
Buck was already a source of consternation for the MIT legal department by the time of his death. The university was locked in a legal battle with IBM and RCA over the patent for magnetic core memories—which eventually yielded $25 million in license payments. Any attempt to enforce unfiled patents on the cryotron, based on Buck’s lab books, would have seen MIT open litigation with IBM on a new front. Buck was a central figure in the magnetic cores case.
The basic argument was that he had been too liberal in sharing information with other institutions who were part of the industrial cooperation agreement overseen by the government. A young engineer at RCA then exploited his generosity, and filed a patent based on what he knew of MIT’s work. It may seem obscure now, but at the time it was one of the biggest patent disputes in history.
JAN RAJCHMAN CAME to America in 1935 as a twenty-four-year-old graduate student. He was born in London to Polish parents, but grew up mostly in Geneva, where his father worked for the League of Nations. After securing his degree in electrical engineering, he wanted to get involved in the newest technology. That meant coming to the United States.
“I was fascinated by electronics, which was the great new field at the time, and a field far more advanced in America than anywhere else,” Rajchman told an interviewer for the Smithsonian Institution in 1970. “Moreover, there was the depression. It was exceedingly difficult to get a position in Europe, particularly for somebody who was not a citizen of the country where he resided, which was my case, since I was then a Polish citizen residing in Switzerland. On the other hand fortunately America had the tradition of accepting immigrants from all over the world, even though there was great depression in America, too. But still, immigrant or no immigrant, everyone was on the same footing as far as getting a job.”
Rajchman did a summer course at MIT partly to dust off his English—he had not used it since the age of seven, when his family left London. In the autumn he was hired by RCA, which was designing television sets and working on radar equipment, to work in its research lab.
By the time World War II broke out, RCA was starting to dabble with primitive computing components, mostly designed by Rajchman. He was then seconded to the University of Pennsylvania to work on the ENIAC project, where he met many of America’s other computing pioneers.
Rajchman never lost touch with MIT, and was regularly on campus wining and dining various professors on his RCA expense account. He became friendly with Buck and Papian. Like the MIT team, Rajchman was also working on computer memory—in league with the Institute for Advanced Study at Princeton University. He had invented a computer memory called the Selectron, another variation on the TV tube design that was being sold in RCA’s machines.
It was technically brilliant, apart from the fact that it kept breaking down. The first RCA computer had seven of these tubes. According to some of the experts who used the machines, one of the seven tubes usually had to be repaired every twenty minutes or so.
Along with every other computer designer in the United States, Rajchman had been part of the information flow around using magnets to store data. He was working on an extremely similar design to that of the MIT lab. He had also built his own oven to make tiny doughnut-shaped magnets.
“During our work on the Selectron I thought of the core memory in a way,” Rajchman told the Smithsonian. “But the fact is, it’s hard for me to imagine the day when I hadn’t thought of the core memory. I thought about it for years before writing anything down.”
Rajchman filed a patent for magnetic core memory in September 1950. It was May of the following year before Forrester filed his patent—which included his idea of three-dimensional computer memory. Forrester’s patent filing was based on the paper that had been sent to Eachus, but souped up with the work that Buck and Papian had done in the lab.
Both MIT and RCA continued to improve their technology, and to share ideas, as they were obliged to do under the cooperation agreements. Buck’s lab notebooks record one meeting with Rajchman in October 1951 where he was shown a switch developed by the RCA lab. It was remarkably similar to something designed by Ken Olsen, Buck’s close friend and colleague at MIT, who would go on to become one of the first captains of industry to emerge from America’s booming new computer business as the founder of Digital Equipment Corporation.
“The switch is his own invention,” wrote Buck in his lab books, of his conversation with Rajchman. “And dates to ‘the weekend before my last visit’ to his laboratory.”
It’s hard to detect from the lab book entry if Buck suspected Rajchman had ripped off the work that had been done in the MIT lab. Soon there would be no doubts in the minds of the MIT leadership, however.
Executives at RCA often wrote letters to Buck asking for details of the latest projects. As one of the big corporate supporters of the industrial cooperation agreement, it was entitled to do so.
“We are specifically interested in circuit operation and operating conditions of the Whirlwind flip-flop, gates, matrices and buffer amplifiers,” wrote Lowell Bensky, an RCA executive who had previously worked in the MIT lab in one letter to Buck in December 1951.
Shortly after requesting the information from Buck’s lab, Rajchman filed a second patent that included some of the developments that had been perfected by Buck and Ken Olsen. According to Bensky, this was no coincidence.
“Jan Rajchman came up under this [industrial cooperation] program and went up to MIT to talk to Forrester,” explains Bensky. “Forrester had this idea for core memory. All other types of memory were subject to failure at the time. So Rajchman came up to talk to Forrester and, according to the rules of the cooperation agreement, Forrester told him everything he was doing. Rajchman then went back to RCA and wrote up a patent based on what Forrester had just told him. And the RCA lawyers went ahead and filed that patent. When MIT found out that this was what had happened, they weren’t too pleased. MIT sued RCA—and they won, of course.”
The legal spat was not quite as straightforward as Bensky suggests. At the time the patent scam was uncovered, nobody quite appreciated the value of the invention at the center of the dispute. During the years of argument that followed, however, the sums at stake got bigger and bigger.
Magnetic cores were clearly a better technology than the valves, tubes, and drums that had gone before. By the time the magnetic memories were commercially available, IBM had already built its first big fleet of machines using the older and less reliable vacuum tube technology; these first generation IBM machines were already installed in banks, insurance companies, and government departments on long-term lease agreements.
As Papian wrote in a memo to Forrester in July 1960, while they gathered evidence for the case,
IBM dragged their collective feet for some time on the question of magnetic-core storage, and were not seriously damaged by such foolishness only because of their dominant business position in the field and a talent for “product-design” which enabled them to put good units into their machine lines in a hurry once they made up their minds. I recall receiving an unexpected tirade on the subject from the manager, or assistant manager, of the component development side of IBM’s research division in Poughkeepsie sometime around late 1953 or early 1954. The gist of said gent’s remarks was that IBM’s machines were doing fine with their storage tubes, that the customers were “foolishly” demanding core storage because of our (MIT’s) ridiculous attitude that core storage was a superior technique, and that he and his colleagues were being caught in the crossfire. When we tried, gently enough, to point out to him that core storage was simple, superior, and satisfactorily tested, it merely drove him closer to apoplexy.
It was August 1954 before IBM successfully tested its first magnetic core memory, which would be used in its XD-1 machine. Soon, by popular demand, the vacuum tube memories were being ripped out of machines across America to be replaced by a much smaller box of magnetic cores. The difference in speed and size was so noticeable that everyone wanted to have the new machines.
Big corporations did not buy computers in these early years of the industry; they leased them. So it fell to the likes of IBM and Bendix Systems to push through the upgrade to the new memory system. They did so, even though the patent litigation between MIT and RCA over who owned the design to this memory system rumbled on in the background.
Just about every computer in America was soon using magnetic core memories. It would be 1968, some fourteen years after IBM started dabbling with the technology, before the issues were finally resolved.
The dispute was not only about who invented the new memory first but also about all the modifications that had been made along the way, each of which had been patented individually. Most of those modifications were not designed by Forrester, but by Buck, Olsen, Papian, and the other researchers working on the project. The switch designed by Olsen that Buck had seen copied in Rajchman’s lab was dragged into the dispute, as were other refinements to the design. Patents were filed and refiled to clean up the arguments. One of Buck’s inventions was split into a separate patent filing, seemingly to clear up one aspect of the legal argument.
Meanwhile, the organization responsible for commercializing MIT’s work remained doggedly focused on the enforcement of Forrester’s memory patents for magnetic cores. Research Corporation, as the business was called, patiently awaited its payday from an expected settlement. The organization sent annual reports to MIT’s inventors detailing the income and expenditure on their patents.
For years the income line read “none” while the expenses grew ever higher. The only commentary on the magnetic cores and the affiliated patents was that “no effort will be made to license his patent application until such time as the Patent Office indicates the scope of the claims that may be allowed.”
With every year that passed, the potential payout grew bigger: by the early 1960s the technology was everywhere. IBM was worried; it was dominating the market with a disputed technology. It, too, had started to claim that it played a big role in developing the magnetic core memory technology in the first place in an attempt to chip away at the size of the bill it would inevitably have to pay to settle the dispute.
A crunch meeting between IBM and MIT on January 26, 1961, opened the door to a deal. Minutes of the showdown also reveal how the murky, fudged divide between the private and public sectors was still a problem.
James Birkenstock, one of the top advisers to IBM president Thomas J. Watson Jr., was the burgeoning computer giant’s lead negotiator. He was up to his neck in IBM’s work with the military. He mapped out the economics of the business for MIT’s benefit; about 95 percent of IBM’s computers were leased, he explained. The 5 percent who bought their own machine paid a price equivalent to fifty months’ rental. Many customers purchased some parts, and leased the others.
Watson promised to send MIT a list of every machine that had been sold or leased, and to whom—including the US government. This was an important wrinkle in the talks: under the industrial cooperation agreement the government did not have to pay royalties on patents, although it did have to pay for the equipment.
“Mr. Birkenstock stated that it is sometimes difficult to know whether a particular machine is used by the government or a private organization,” the minutes of the meeting note. “For example, a machine used by Rand Corporation may be at work on both Government and private contracts; or a machine used by Pan-American [Airlines] may be used in connection with missile testing at Cape Canaveral. Mr. Birkenstock indicated, incidentally, that IBM’s charges are the same on Government and private machines.”
The MIT negotiators pressed for details on what IBM thought it had done in the development of the magnetic core memory. Birkenstock pointed to IBM’s role in the Whirlwind project and the military computer it subsequently developed to run the defense shield. He denied the MIT line suggesting that IBM had been trying to stifle the magnetic core technology. IBM had been working flat out on magnetic core memory, Birkenstock claimed, but “while all of MIT’s developments were publicized, IBM’s developments were kept secret, as is usually the practice in private industry.”
Birkenstock went on to claim that MIT’s technologies were only of interest to the military because he and his colleagues had pushed for the MIT work to be used rather than a rival technology developed by the University of Michigan.
“Mr. Birkenstock also insisted that computer development did not start at MIT, but that it was IBM who built the first computer, the SSEC [a successor to the Harvard Mark I], which was given to Harvard,” the meeting minutes note. “The SSEC is a systems patent covering every computer in existence today.”
Although Birkenstock claimed at the start of the meeting that he did not want to discuss a settlement, he soon started talking numbers. The MIT lawyers proposed that IBM pay two cents for every magnetic core it used.
According to the minutes of the meeting, “Mr. Birkenstock emphasized that IBM’s operations in large-scale computers have been running at a loss since their inception. Later in the discussion, however, he said that a royalty of 2 cents a core would mean that some computers that are presently being sold at a profit would have to be sold at a loss.”
Birkenstock explained that MIT had completely misunderstood the costs of IBM’s computer business. Component costs were falling quickly, so the two-cent royalty would be equivalent to 4 percent of sales. The total cost of the machine was not just about parts, however. IBM sold electric typewriters at a markup of three times manufactured cost, and computers at ten times the manufactured cost, he explained. Yet a “mark-up of such magnitude is made necessary by the high cost of selling and of customer assistance.”
A royalty payment equivalent to 4 percent of sales would have “serious consequences” for IBM, Birkenstock said. But it would be “even more drastic for other members of the industry, possibly forcing them out of business.” He then proposed an alternative settlement, based on a series of complicated sums. It worked out at a flat fee of between $1.5 million and $1.6 million to cover all past and future infringements of MIT’s patent.
It would take another seven years of negotiations to hammer out the final deal between IBM and MIT, yet the broad structure did not really change. The talks were held up for several years until the original patent dispute between MIT and RCA was resolved. Eventually the argument was settled thanks to a paper written in 1950 by Forrester—the same paper that Buck had sent to Joseph Eachus at Seesaw days after first arriving at MIT. Irrespective of who came up with the idea first, it was clear that Forrester had written it down before anyone else. As a result, IBM decided to settle with MIT and Rajchman lost out.
The $1.5 million that Birkenstock had initially proposed ballooned by 900 percent, to more than $13 million in the final settlement, according to legal papers released by MIT in the research for this book. The settlement forced every other computer manufacturer that had been using the technology to reach their own settlement. In total, MIT received more than $25 million from legal settlements over the patent, equivalent to about $176 million in today’s money.
As the sole inventor listed on the magnetic core memory patents, Jay Forrester was suddenly a rich man. MIT had a system where the inventor got a 12 percent cut of the royalty payments. As a result Forrester got about $3 million over the years, or about $21 million in today’s money.
Ken Olsen got what he described in a letter to the Buck family years later as a “nominal sum” for his work on the switch. That “nominal sum” still ran to tens of thousands of dollars, however. A letter sent to Forrester on February 26, 1965, from Paul Cusick, MIT’s comptroller, details the first payments made to the inventors of the magnetic core memory patent. MIT had received a check for $2.8 million, so Forrester was being sent $323,373—described as 12 percent of what he was owed, “less $10,707 which was delivered to Kenneth H. Olsen.”
The formula by which Olsen’s share was calculated is not possible to understand from the documents in MIT archives; a side deal of some kind appears to have been negotiated. It equates to almost $80,000 in today’s money for that first installment alone.
Olsen started to receive these monies at the time he was setting up Digital Equipment Corporation, which went on to become one of America’s biggest manufacturers of office microcomputers. It employed 140,000 people at its peak, and was eventually bought by Compaq in 1998 for $9.6 billion—at the time, the biggest deal ever in the world of computers.
According to sources familiar with the settlement, Olsen also gained the right to use the magnetic core patent for free as part of his negotiations with MIT in regard to the legal settlement in recognition of the fact that he had done a great deal of the work on the technology.
Although Buck played a similar role in developing the patents, he received nothing for his work. He was named in the documents and his patents helped to swing the deal, but neither Buck nor his family received a penny from the settlement.
Internal correspondence between Forrester and the MIT team running the legal case suggests that Buck may have been blamed for sharing too much information about the secrets of Project Whirlwind. Forrester knew that Buck was still in regular contact with his old colleagues at Seesaw. As the case started to reach the crunch phase of negotiation, he seems to have looked to Buck as the possible source of a leak.
“It was his [Buck’s] duty to report to them information that might be of interest to them which was happening at the Digital Computer Laboratory, and he also made extensive trips to other organizations and lectured on the computers and components indicating what we were doing at the Digital Computer Laboratory,” Forrester wrote before the case had been settled. “Buck’s name should be added to the list of people whose telephone calls and travel vouchers are being traced by the accounting office. We should try to reconstruct all the organizations he went to during 1950 and the first half of 1951. In particular, his reports to the National Security Agency may contain very valuable information.”
By the time IBM and MIT put pen to paper on their agreement, Buck had been dead for nine years. Birkenstock, the IBM negotiator, slipped in an extra condition late in the discussions: IBM also wanted its hands on “the Buck patent.”
The patent in question was not for the light gun, or any of the modifications to the magnetic core memory patent; it was the technology that had brought Dudley Buck international fame and recognition. IBM wanted the right to use the cryotron—and it was duly granted. At the time, IBM still believed the cryotron would be the future of the computer.
Project Lightning ended up being about a lot more than just cryotrons. It always suffered a stigma from the fact that it did not produce a single machine that everyone could point to when senior figures in government asked where all the millions of dollars had gone. Yet it has been credited in several papers with providing the impetus for new ideas.
It was Project Lightning that concentrated the industry’s minds on the idea that quicker circuits led to better computers. That in turn led to Moore’s law, the thesis posited by Intel cofounder Gordon Moore that microchips double in speed every two years. Although there is no physical or scientific basis for Moore’s law, it is an observation that has held true since it was coined, with those advances in speed leading to the extremely mobile electronics of the twenty-first century.
Snyder’s official history of NSA computing projects cites Project Lightning in glowing terms for this very reason. It mentions how the cryotron “proved not to scale to high speed operation as had been hoped.” The detailed explanation of how the cryotron was used and what went wrong with it remains classified. It seems that it never was used as a missile guidance system, in spite of the time that was spent on the idea; the semiconductor took that crown.
Yet the cryotron retained a hard-core group of fans among the senior ranks of the American science establishment.
IBM spent years—and something of the order of $250 million—working on superconducting microchips. The cryotron evolved into a device called the Josephson junction by the mid-1960s. In 1987, scientists at IBM won international acclaim for creating a superconductor that could switch between states at much higher temperatures—opening the door to superconductors that could operate at room temperature.
In February 2012 the company’s Watson Research Lab unveiled a superconducting quantum computer. Although the theory behind it relies on the mad world of quantum mechanics, the core materials used in the chip are silicon, aluminum, and niobium—the same materials Buck had been experimenting with fifty-three years earlier. It runs on modified cryotrons.
Intriguingly, in the fallow years when American researchers paid little heed to superconducting microchip research, Russian researchers made significant strides. The technology became central to the development of the Soviet Union’s most advanced computer chips through to the fall of the Berlin Wall in 1989.
As David Brock of the Computer History Museum explains,
The Russians kept up with superconducting electronics. They were consistently into it in the same way that the NSA was into it. IBM tried to build this computer with the next generation of cryotrons, which were originally called tunneling cryotrons, and now they are called Josephson junctions. IBM had this gigantic project; that failed, not really technically, but IBM did not see it as commercially viable, so it died.
Until the recent interest in quantum computing—much of which is based on these tunneling cryotrons—the technology had lain dormant in the US. But the Russians always kept going with it. In fact, a lot of the new stuff that’s in superconducting electronics and the superconducting approach to quantum computing are from Russians who got out after the fall of the Berlin Wall. I have no doubt that the Soviets were really interested in cryotrons, because they kept with it.
In 1985, researchers at Moscow State University outlined a theory for a new superconducting chip it called rapid single flux quantum. It was a faster, more energy-efficient interpretation of the Josephson junction—that is, another modified cryotron.
By 1997 Moscow State University had formed a partnership with Bell Laboratories (with backing from the NSA) named the Hybrid Technology Multi-Threaded project, which was tasked with finding a replacement for silicon to produce petaflop-paced supercomputers. Four decades after Dudley Buck’s death, the NSA still believed the cryotrons could beat a silicon transistor. That specific project ended in 2000.
By the early 2010s, however, NSA researcher Marc Manheimer persuaded the agency to have another look at superconducting supercomputers. Manheimer, based at the NSA’s Laboratory for Physical Sciences, almost immediately next door to NSA headquarters in Maryland, has said that he encountered multiple skeptics due to a “history of failure” with Dudley Buck–inspired technology.
Yet by 2013 Manheimer had won over Intelligence Advanced Research Projects Activity (IARPA), the research department for the intelligence community, to create the Cryogenic Computing Complexity program. He switched agencies to run the program, the budget for which has never been disclosed.
Public disclosures from 2014 show that defense titans Northrop Grumman and Raytheon were awarded a slice of the contract, along with IBM. And much of the work is being conducted in MIT’s Lincoln Laboratory—where Buck began work on Project Whirlwind in 1950.
The NSA and IARPA are not the only agencies to have continued dabbling with Buck’s technology; superconductors remained popular with NASA for years after his death. Wernher von Braun, the German-born mastermind of the V-2 flying bomb who had switched sides after World War II to lead missile and space research in the United States, had something of an obsession with the subject.
The Josephson junction—the successor to the cryotron—was used by NASA for years. Although it was a different invention, Braun insisted on still calling the new switch a cryotron. NASA technical papers suggest that the space agency continued research into these cryotron-like switches until at least the mid-1990s.
Braun wrote in the January 1969 edition of Popular Science magazine about how magnets and superconductors could be used to create a force field that would protect spaceships from solar flares on future missions to Mars. Braun made it quite clear how he thought these future spaceships would find their way:
The propensity of many superconducting materials to “go normal” (lose their superconductivity) in a magnetic field has one useful and redeeming aspect. It permits their employment as contactless switching devices, called cryotrons. A cryotron consists of a thin-film “gate wire” and a “control wire,” both superconductive. Send a current through the control wire, and its magnetic field kills the superconductivity of the gate wire, giving the effect of an on-off switch.
All basic types of electronic computers’ circuits can be built from combinations of these microminiature switching units. The resulting computer, which is kept refrigerated in operation, is reduced to shoebox size and consumes extraordinarily little power—ideal qualities for space use. A cryotron computer and a superconducting gyro and accelerometer could make up a high-precision navigation system to help future astronauts find their way about the solar system.
At a recent conference in Abu Dhabi, a former administrator of NASA was asked over dinner about cryotrons and Josephson junctions. He hesitated, scowled, then barked: “How the hell do you know about that?”