by Yasha Levine
And networks? They existed. But, like the network that tied SAGE together, they were usually highly specialized and built for a particular purpose and function. A network would have to be designed and custom built to fit every new situation.
The way Lick saw it, this was the wrong way to handle the command and control technology problem. What ARPA needed was to develop a universal and standardized computer and networking platform that could be modified with minimal effort to handle just about any task: missile tracking, behavioral studies, databases, voice communication, intelligence analysis, or simple text processing and mail functions. This computer framework would have a few basic underlying components. It would be easy to use and have an intuitive graphical user interface, feature a universal operating system and programs that could be loaded onto it, and, most important, would move away from the calculator mode of computer operation by allowing users to work in real time in the same way people interact with one another. Though this may sound basic and obvious, these kinds of computer tools did not exist in the early 1960s.
“There was the belief in the heads of a number of people—a small number—that people could become very much more effective in their thinking and decision making if they had the support of a computer system, good displays, and so forth, good data bases, computation at your command. It was the kind of image that we were working toward the realization of,” explained Lick in an ARPA report.34 “It really wasn’t a command and control research program. It was an interactive computing program. And my belief was, and still is, you can’t really do command and control outside the framework of such a thing.”
The crude state of computer technology meant that Lick’s goal was still years away, and one thing was for sure: it wouldn’t be invented on its own. Someone had to do the work. As Lick saw it, ARPA’s primary mission was to throw money at engineers who could build the underlying computer components that a modern command and control system required. At a minimum, ARPA would at least get people working on computer projects that pointed in the right direction. Lick saw his job in historical terms. He would use ARPA’s budget and influence to push the computer industry into a new territory, one that aligned with his vision and the needs of the defense establishment.
But first, he wanted to make sure that US intelligence agencies hadn’t secretly developed this kind of interactive computing technology already. “I even went over to the CIA and gave them a pitch,” said Lick. “I had to tell them, ‘Look, I do not know what you’re doing about this. I hope you are doing the following. But let me tell you about what I am doing, and then maybe we can figure some way to talk about what the relations are.’” He also arranged a meeting with reps from the NSA and made the same pitch about the beauty of a universal, easy-to-use computer platform. Neither agency was working on interactive computing, but they sure wanted to get their hands on it—“the NSA, they really needed what I wanted,” he recalled in an interview years later.35 Indeed, intelligence agencies were among the first users of the tools ARPA’s command and control program produced just a few years later.
ARPA’s initial Command and Control Research budget was $10 million. Lick spread that cash through his personal and professional networks in the military-academic-contractor world. He bankrolled projects on interactive computing and time sharing, graphical interface design, networking, and artificial intelligence at MIT, UC Berkeley, UCLA, Harvard, Carnegie Mellon University, Stanford, and the RAND Corporation. At MIT, Lick set up one of his biggest and most important initiatives: Project MAC, short for Machine-Aided Cognition, which evolved into a sophisticated interactive computer environment complete with email, bulletin boards, and multiplayer video games. MIT’s Project MAC spawned the first crop of “hackers,” ARPA contractors who tinkered with these giant computers in their free time.
At the Stanford Research Institute, which was also doing ARPA contract work on chemical warfare in Vietnam, Lick funded Douglas C. Engelbart’s Augmentation Research Center. This team became legendary in computer circles. It developed hypertext links, multiuser real-time word processing, video conferencing, and, most notably, the computer mouse. Lick also jumpstarted a whole range of networking projects, efforts that would lead directly to the creation of the Internet. One of these was a $1.5 million joint UCLA–UC Berkeley initiative to develop software and hardware for a network that connected multiple computers to multiple users.36 As a funding proposal explained, this research would be used directly to improve military networks, including the National Military Command System, which was then a new communication system linking the military to the president.37
Lick worked hard and fast, and his efforts at ARPA were remarkable. Companies like General Electric and IBM did not initially accept his ideas about interactive computing. But with his tenacity and ARPA’s funding, his vision gained traction and popularity and ultimately changed the direction of the computer industry. His tenure at ARPA achieved something else as well: computer science became more than just a subdivision of electrical engineering; it developed into a proper field of study of its own.38 The long-term research contracts the ARPA Command and Control Research division handed out to research teams helped seed the creation of independent computer science departments in universities across the country and tied them closely, through funding and personnel, to the US military establishment.
Networking: The Dark Side
Computer history buffs consider Lick one of the most important personalities in the development of computer science and the Internet. A five-hundred-page biography, called The Dream Machine by M. Mitchell Waldrop, chronicles Lick’s life and work. What almost never gets reported, but what comes through the pages and pages of released and declassified government files covering Lick’s tenure at ARPA, is just how much his computer research efforts were infused with the agency’s greater counterinsurgency mission.
Lick died in 1990, a few months shy of turning seventy-five. In interviews, he had made sure to distance his efforts at ARPA from the agency’s less wholesome work fighting insurgencies. “There was a kind of a cloak and dagger part of it,” he recalled in a 1988 interview.39 “There was a fellow named Bill Godel who, it seemed to me, was always trying to get control over what I was doing. I could never tell what he was doing, so that part made me nervous. I had one project that I wasn’t cleared deeply enough to know what was, and that made me nervous.” He readily conceded that he knew something shady was cooking at ARPA, and hinted that he took part in some of it, but claimed that he resisted attempts to involve his command and control project in unsavory Vietnam counterinsurgency efforts. “I sort of stayed out of that as best I could,” he explained.
The truth is a bit stickier.
Lick’s job was to develop the underlying computer and networking technology necessary to fight modern wars. Naturally, this applied to counterinsurgency in a very general way. But his work was also much more specific and direct.
For instance, documents show that in March 1962, he attended an influential US Army symposium that convened in Washington, DC, to discuss how behavioral science and computer technology could be used to better wage “limited war” and counterinsurgency. There, Lick was part of a working group dedicated to crafting a US Army counterinsurgency research program that could meet a “multidimensional Communist challenge—in paramilitary warfare, in psychological warfare, and in the conventional and nuclear field.”40 The symposium took place just as Lick was starting his job as head of ARPA’s Behavioral Science and Command and Control Research divisions. Going forward, his work at ARPA was part of the military’s larger counterinsurgency efforts and directly overlapped with William Godel’s Project Agile.41
Naturally, many of ARPA’s programs in Southeast Asia—from remote-control drones to electronic sensor fences and large-scale human intelligence gathering—were all tied in one way or another to data collection and communication, and they ultimately depended on computer technology to organize and automate these tasks. They necessitated tools tha
t could ingest data on people and political movements, compile searchable databases, tie in radio and satellite communications, build models, predict human behavior, and share data quickly and efficiently over great distances between different agencies. Building the underlying technology that could power all newfangled communication platforms was Lick’s job. He certainly never shied away from steering research toward counterinsurgency applications. A glance at the contracts from those days shows him directing funds to projects that used computers for everything from studying and predicting the behavior of people and political systems to modeling human cognitive processes and developing simulations that predicted “the behavior of international systems.”42 Records show that as early as 1963, Lick’s Command and Control Research division was sharing and intermingling funds with William Godel’s Project Agile.43
Indeed, even as Lick started at ARPA, Project Agile was deploying data-driven counterinsurgency initiatives in the field. One of the earliest took place between 1962 and 1963 at ARPA’s Combat Development Test Center in Thailand, on the outskirts of Bangkok. It was called Anthropometric Survey of the Royal Thai Armed Forces. On the surface, it was a benign study that sought to measure the body size of several thousand Thai military personnel to aid in the design of equipment and uniforms. It collected fifty-two different data points, everything from sitting height to buttock–knee length to crotch–thigh circumference and seven different measurements of the face and head.
The survey’s data points had the unpleasant feel of a eugenic study, but the physical measurements were just the surface level of the study. The deeper purpose was rooted in prediction and control.44 “Thai participants were also asked a bevy of personal questions—not just where and when they were born, but who their ancestors were, what their religion was, and what they thought of the king of Thailand,” explains Annie Jacobsen in The Pentagon’s Brain. These questions were at the heart of the study’s true goal: to create a computer profile of each Thai serviceman and then use it to test predictive models. “ARPA wanted to create a prototype showing how it could monitor third world armies for future use. The information would be saved in computers stored in a secure military facility. In 1962 Thailand was a relatively stable country, but it was surrounded by insurgency and unrest on all sides. If Thailand were to become a battle zone, ARPA would have information on Thai soldiers, each of whom could be tracked. Information—like who deserted the Thai army and became an enemy combatant—could be ascertained. Using computer models, ARPA could create algorithms describing human behavior in remote areas.”45
The link between counterinsurgency and computers is not that surprising. The first rudimentary computer technology was developed in the United States almost a century before the Vietnam War to count, categorize, and study masses of people. In the late 1880s, an American by the name of Herman Hollerith invented a tabulation machine under contract with the US government to speed the process of counting people for the US Census. Because of a huge immigration influx, the census had become so unwieldy that it took a full decade to finish the count by hand.
Hollerith came up with an elegant electromechanical solution, a contraption that would later become the backbone of International Business Machines, or IBM, the oldest computer company in the world. His design broke down the process of automatic data calculation into two general steps. First, data were digitized, that is, converted into a format that could be understood by a machine, via a series of holes punched in a piece of paper. The second step involved feeding this paper into an apparatus containing electrical pins that tabulated and sorted the punch cards on the basis of position and arrangement of the punched holes. Hollerith initially thought to record the information on a long strip of paper, like a ticker tape. But he quickly abandoned the idea because it made it too difficult to locate and isolate individual records—in a census the machine would process hundreds of thousands and even millions of individuals. “The trouble was that if, for example, you wanted statistics regarding Chinamen, you’d have to run miles of paper to count a few Chinamen,” Hollerith explained.46
So, he went with a different idea: each person would be represented by a separate punch card. The inspiration came from an observation he made on a train. To prevent people from passing around and reusing train tickets, conductors punched out a passenger’s description on a little slip of paper: height, type of hairstyle, eye color, and nose type. It was an elegant and powerful solution. Each person had their own card—and each card had a standardized pattern of holes that corresponded to information collected by the census takers. Each card would encode a person’s attributes: age, sex, religion, occupation, place of birth, marital status, criminal history. Once a clerk transferred the data from a census form onto a punch card, the cards would be fed into a machine that could count and arrange them in all sorts of ways. It could provide aggregate totals for each category or find and isolate groups of people in specific categories. Any trait—nationality, employment status, disability—could be singled out and sorted quickly. Hollerith described his system as making “a punch photograph of each person.” And, indeed, it did: a first-generation digital dossier of people and their lives.
Used to count the census in 1890, Hollerith’s tabulators were a huge success, cutting the time it took to crunch the numbers from years to months. The machines also lent census trackers the ability to slice, dice, and mine the data in ways that had never been possible; for example, to find a particular person or group of people—say, Americans with at least one Japanese parent in California or all orphans living in New York with a felony. This kind of fine-grained analysis on a mass scale was unprecedented. Overnight, Hollerith’s tabulators transformed census taking from a simple count into something very different—something that approached an early form of mass surveillance.
Newton Dexter North, a wool industry lobbyist chosen to head the 1900 census, was astounded by the ability of Hollerith’s tabulators to so precisely tabulate racial data. Like many upper-class Americans of his day, North worried that the massive influx of immigrants from Europe was destroying the fabric of American society, causing social and political unrest, and threatening the nation’s racial purity.47 This fear of immigration would become intertwined with anticommunist hysteria, leading to repression of workers and labor unions across the country. North saw statisticians like himself as technocratic soldiers: America’s last line of defense against a foreign corrupting influence. And he saw the tabulator machine as their most powerful weapon. “This immigration is profoundly affecting our civilization, our institutions, our habits and our ideals. It has transplanted here alien tongues, alien religions, and alien theories of government; it has been a powerful influence in the rapid disappearance of the Puritanical outlook upon life,” North warned, but he heaped praise on Hollerith’s newfangled computation device. “I cannot detain the reader with a statement of the correlation of the data of individual elements of the population, in combination with other data, beyond the reach of hand tabulation, which this invention opened up,” he explained. “Without it we could never hope to lay bare all the truth we must have, if we are to cope successfully with the problems growing out of the heterogeneous commingling of races which our defective immigration laws are forcing upon us.”48
Two decades after its debut, Hollerith tabulation technology was absorbed into IBM. Improved and refined over the years, the machines became a runaway hit with businesses and government. They were used extensively by the US military during World War II to keep an up-to-date tally of troop numbers and were even dragged onshore during the invasion of Normandy. They were also used to process the internment of Japanese Americans during the war. And, after President Franklin Delano Roosevelt created the Social Society system, IBM and its tabulators functioned as a de facto privatized arm that did all the processing and accounting for America’s pension system.49 Perhaps most infamously, IBM’s tabulator machines were employed by Nazi Germany to run death labor camps and to institute a system of racia
l surveillance by enabling the regime to comb genealogical data to root out people with traces of Jewish blood.50
Willy Heidinger, head of IBM operations in Germany and a devout member of the Nazi Party, knew the part he played, with the help of IBM tabulators, in studying a sick German people and helping Adolf Hitler provide the cure: “We are very much like the physician, in that we dissect, cell by cell, the German cultural body. We report every individual characteristic… on a little card,” he said in a fiery speech dedicating a new IBM factory in Berlin. “We are proud that we may assist in such a task, a task that provides our nation’s Physician with the material he needs for his examinations. Our Physician can then determine whether the calculated values are in harmony with the health of our people. It also means that if such is not the case, our Physician can take corrective procedures to correct the sick circumstances.… Hail to our German people and der Fuhrer!”51
Nazi Germany’s use of IBM technology is an extreme example, but it underscores the connection between the development of early computer technology and the study and management of large groups of people. IBM tabulators remained in operation through the 1980s. Indeed, until J. C. R. Licklider and ARPA developed interactive computing systems, tabulators and punch cards were the principal means by which militaries, government agencies, and corporations wrote programs and worked with complex data sets.
There is no doubt that Licklider’s computer research at ARPA was intimately bound to the agency’s expanding counterinsurgency mission.52 But in internal discussions with his ARPA contractors—engineers and social scientists at major universities across the country—Lick sought to deemphasize the military applications of his command and control project, instead shifting the focus to the need to build productivity-boosting computer technology for his civilian collaborators and their colleagues.