Surveillance Valley

by Yasha Levine

  In a letter to his contractors, Lick wrote:

  The fact is, as I see it, that the military greatly needs solutions to many or most of the problems that will arise if we tried to make good use of the facilities that are coming into existence. I am hoping that there will be, in our individual efforts, enough evident advantage in cooperative programming and operation to lead us to solve the problems and, thus, to bring into being the technology that the military needs. When problems arise clearly in the military context and seem not to appear in the research context, then ARPA can take steps to handle them on an ad hoc basis. As I say, however, hopefully, many of the problems will be essentially the same, and essentially as important, in the research context as in the military context.53

  On a fundamental level, the computer technology required to power active military operations was no different from the tech scientists and researchers used to do their work. Collaboration, real-time collection and sharing of data, predictive modeling, image analysis, natural language processing, intuitive controls and displays, and computer graphics—if the tools developed by ARPA contractors worked for them and their academic buddies, they would also work for the military with only slight modifications. Today’s military takes this for granted: computer technology is always “dual use,” to be used in both commercial and military applications. Deemphasizing ARPA’s military purpose had the benefit of boosting morale among computer scientists, who were more eager to work on the technology if they believed it wasn’t going to be used to bomb people.54

  Two years into his job at ARPA, Lick began to view the various computing projects he had seeded all over the country—from UCLA to Stanford and MIT—as parts of a larger connected unit: computer “thinking centers” that at some point in the near future would be netted together into a single, unified, distributed computing machine. It mirrored the vision of a networked society he had outlined in 1960: first, you connect the powerful computers via a high-bandwidth network. Then you connect users to these computers with telephone lines, satellite dishes, or radio signals—whichever technology was best suited to their particular needs. It would not matter whether people logged in from home, work, a jeep crawling through the jungles of Vietnam, or a stealth bomber flying ten miles above the Soviet Union. “In such a system, the speed of the computers would be balanced, and the cost of the gigantic memories and the sophisticated programs would be divided by the number of users,” he had written. In 1963, four years after publishing that paper, Lick began coyly referring to this idea as the “Great Intergalactic Network.” Fundamentally, his vision for a distributed interactive computing network is not very different from what the Internet looks like today.55

  In 1964, two years after coming to ARPA, Lick decided that he had fulfilled his mission of getting the agency’s Command and Control Research program up and running. He moved his family to Westchester County, New York, to start a cushy gig running a research division at IBM.56 Younger and more energetic men would have to finish the job he started.

  The ARPANET

  Lawrence Roberts was twenty-nine years old when he reported to duty at ARPA’s Command and Control Research division inside the Pentagon. The year was 1966, and he was hired for a big and important job: to make Lick’s Great Intergalactic Network a reality.

  Everything was in place. ARPA had a range of functional overlapping interactive computer projects spread across the country, including at the following centers:

  MIT’s Artificial Intelligence Laboratory

  MIT’s Project MAC

  Stanford’s Artificial Intelligence Laboratory

  Stanford’s Research Institute

  Carnegie Mellon University

  University of California, Irvine

  University of California, Los Angeles

  University of California, Berkeley

  University of California, Santa Barbara

  RAND Corporation

  Utah University

  It was time to wire all these computer centers together and have them function as one unit. It would be called the ARPANET.

  Roberts came from the MIT Lincoln Laboratory, where he had been working on graphics and computer communication systems. Some of his colleagues found the strict atmosphere there stifling. In fact, two of them left in a huff because of the lab’s “no pets” policy. “They wanted to bring a cat into the lab. Lincoln would not allow them to bring a cat in. And they decided that was unfair; they would go somewhere where cats were tolerated,” he recalled, noting wryly that the cats were not for companionship but for gruesome experimentation. “It was really a fight over having that connection with the brain electrodes and all of that. Lincoln just did not want anything to do with it.”57

  But Roberts had no such problem. He had a broad forehead, big, floppy earlobes, and a stern but calm and measured way of talking. He was a math and theory kind of guy. He thrived at Lincoln Lab, working on moon-shot algorithms, image compression, and data network design. He knew Lick and was inspired by his vision of a universal network that could net all sorts of systems together. Indeed, Roberts was an efficient networker. “Within a few weeks, he had the place—one of the world’s largest, most labyrinthine buildings—memorized. Getting around the building was complicated by the fact that certain hallways were blocked off as classified areas. Roberts obtained a stopwatch and began timing various routes to his frequent destinations,” write Katie Hafner and Matthew Lyon in their upbeat and zany book about the creation of the Internet, Where Wizards Stay Up Late.58 Inside the Pentagon, people started calling the most efficient path between two points “Larry’s Route.”

  Roberts liked building networks, just not the social kind. He was reserved and extremely socially averse. None of his coworkers, not even the ones closest to him, knew much about him or anything about his personal life. He was obsessed with efficiency and was really into speed reading, studying and improving his technique to the point where he could read thirty thousand words a minute. “He’d pick up a paperback and be through with it in ten minutes. It was typical Larry,” one of his friends recalled.

  Roberts’s task was daunting: connect all of ARPA’s far-flung interactive computer projects—with computers made by a half dozen different companies, including a one-of-a-kind ILLIAC supercomputer—into one network. “Almost every conceivable item of computer hardware and software will be in the network. This is the greatest challenge of the system, as well as its greatest ultimate value,” said Roberts.59

  Not long after arriving at ARPA, he convened a series of meetings with a core group of contractors and several outside advisers to hash out the design. The sessions brought together a mix of ideas and people. One of the most important was Paul Baran, who had worked at RAND designing communication systems for the air force that could survive a nuclear attack.60 Over time, the group came up with a design: key to the network would be what Roberts called interface message processors, or IMPs. These were dedicated computers that would form the connective tissue of the distributed network. Connected by telephone lines leased from AT&T, they would send and receive data, check for errors, and ensure that data successfully reached the destination. If part of the network went down, the IMPs would attempt to retransmit the information using a different pathway. IMPs were the generic gateways to ARPA’s network, functioning independently of the computers that used them. Different makes and models of computers did not need to be designed to understand each other—all they needed to do was communicate with the IMPs. In a way, IMPs were the first Internet routers.

  Finally, in July 1968, Roberts put out a contract request to over a hundred computer companies and military contractors. Bids came back from some of the biggest names in the business: both IBM and Raytheon were interested, but the contract was ultimately awarded to an influential early computer research firm in Cambridge, Massachusetts, called Bolt, Beranek and Newman, where J. C. R. Licklider was a senior executive.61

  The very first ARPANET node, powered by the IMPs, went live on Octob
er 29, 1969, linking Stanford to UCLA.62 The first attempt to connect barely worked and dropped after a few seconds, but by the next month, connections to UC Santa Barbara and University of Utah were also made. Six months later, seven more nodes became operational. By the end of 1971, more than fifteen nodes existed. And the network kept growing.63

  In October 1972, a full demonstration of the ARPANET was carried out at the first International Conference on Computer Communications in Washington, DC. It astounded people. ARPA contractors fit out a hall with dozens of computer terminals that could access computers across the country and even a link in Paris. Software available for demonstration included an air traffic simulation program, weather and meteorological models, chess programs, database systems, and even a robotic psychiatrist program called Eliza that provided mock counseling. Engineers ran around like children at an amusement park, overwhelmed by how all the different parts flawlessly fit together and worked as one interactive machine.64

  “It was difficult for many experienced professionals at that time to accept the fact that a collection of computers, wide-hand circuits, and minicomputer switching nodes—pieces of equipment totaling well over a hundred—could all function together reliably, but the ARPANET demonstration lasted for three days and clearly displayed its reliable operation in public,” Roberts recalled. “The network provided ultra-reliable service to thousands of attendees during the entire length of the conference.”65

  Even so, not everyone was excited by what ARPA was doing.

  ”The Octoputer: Serves the Ruling Class”

  September 26, 1969, was a mild fall day at Harvard University. But all was not well. Several hundred angry students gathered on campus and marched on the office of Harvard’s dean. They piled inside and refused to leave. A day earlier, five hundred students had marched through campus, and a small contingent of activists from Students for a Democratic Society had broken into the school’s Office of International Affairs and forced the administrators out onto the street.66 Similar troubles were afoot just across the river at MIT, where students were holding protests and teach-ins.67

  Fliers posted on both campuses railed against “computerized people-manipulation” and “the blatant prostitution of social science for the aims of the war machine.” One leaflet warned: “Until the military–social science complex is eliminated, social scientists will aid the enslavement, rather than the liberation, of mankind.”68

  What exactly were the students protesting?

  The ARPANET.

  Vietnam is the most blatant example of the U.S. attempt to control underdeveloped countries for its own strategic and economic interests. This global policy, that prevents the economic and social developments of the third world, is imperialism.

  In serving these policies, the U.S. government has no qualms about setting up a project that ties together MIT, Harvard, Lincoln Labs, and the entire Cambridge research and development complex.69

  Earlier that year, activists from Students for a Democratic Society got their hands on a confidential ARPA proposal written by none other than J. C. R. Licklider. The document ran to almost a hundred pages and outlined the creation of a joint Harvard-MIT ARPA program that would directly aid the agency’s counterinsurgency mission. It was called the Cambridge Project. Once complete, it would allow any intelligence analyst or military planner connected to the ARPANET to upload dossiers, financial transactions, opinion surveys, welfare rolls, criminal record histories, and any other kind of data and to analyze them in all sorts of sophisticated ways: sifting through reams of information to generate predictive models, mapping out social relationships, and running simulations that could predict human behavior. The project emphasized providing analysts with the power to study third-world countries and left-wing movements.

  Students saw Cambridge Project, and the bigger ARPANET that plugged into it, as a weapon. A pamphlet handed out at the MIT protest explained: “The whole computer set-up and the ARPA computer network will enable the government, for the first time, to consult relevant survey data rapidly enough to be used in policy decisions. The net result of this will be to make Washington’s international policeman more effective in suppressing popular movements around the world. The so-called basic research to be supported by Project CAM will deal with questions like why do peasant movements or student groups become revolutionary. The results of this research will similarly be used to suppress progressive movements.”70 Another booklet featured a mock advertisement that gave a visual representation to these fears. It featured “The Octoputer,” a computer shaped like an octopus that had tentacles reaching into every sector of society. “The Octoputer’s arms are long and strong,” read the mock ad copy. “It sits in the middle of your university, country and reaches helping hands out in all directions. Suddenly your empire works harder. More of your agents use the computer—solving more problems, finding more facts.”71

  To activists, ARPA’s Cambridge Project was part of a networked system of surveillance, political control, and military conquest being quietly assembled by diligent researchers and engineers at college campuses around the country. The college kids had a point.

  The Cambridge Project—also known as Project CAM—was born out of an idea proposed in 1968 by Licklider and his longtime colleague Ithiel de Sola Pool, an MIT political science professor and expert in propaganda and psychological operations.

  As head of ARPA’s Command and Control Research project and Behavioral Sciences program, Lick had seen how the agency struggled with the mountains of data generated by its counterinsurgency initiatives in Southeast Asia. A major goal of his work during his brief stint at ARPA was to jump-start a program that would ultimately build the underlying systems that could make computer-aided counterinsurgency and command and control more efficient: tools that ingest and analyze data, create searchable databases, build predictive models, and allow people to share that information across vast distances. Pool was driven by the same passion.

  Pool, a descendant of a prominent rabbinical family that traced its roots to medieval Spain, was an MIT professor and renowned expert in communications and propaganda theory. Starting in the late 1950s, he ran MIT’s Center for International Studies, a prestigious department for communication studies that was funded by the CIA, and helped set up MIT’s Department of Political Science. He was a hardcore anticommunist and a pioneer in the use of opinion polling and computer modeling for political campaigns. With his expertise, he was tapped to guide the messaging for John F. Kennedy’s 1960 presidential bid, crunching poll numbers and running simulations on issues and voter groups. Pool’s data-driven approach to political campaigning was on the cutting edge of a new wave of electoral technologies that sought to win by pretesting people’s preferences and biases and then calibrating a candidate’s message to fit them. These new targeted messaging tactics, enabled by rudimentary computers, had a lot of fans in Washington and over the next several decades would come to dominate the way politics were done.72 They also inspired fear that America’s political system was being taken over by manipulative technocrats who cared more about the marketing and selling of ideas than they did about what those ideas actually meant.73

  Pool was much more than a campaign pollster; he was also an expert in propaganda and psychological operations and had close ties to ARPA’s counterinsurgency efforts in Southeast Asia, Latin America, and the Soviet Union.74 From 1961 through 1968, his company, the Simulmatics Corporation, worked on ARPA counterinsurgency programs in South Vietnam as part of William Godel’s Project Agile, including a major contract to study and analyze the motivation of captured Vietnamese rebels and to develop strategies to win the allegiance of South Vietnamese peasants. Pool’s work in Vietnam helped further the idea that a purely technical solution could stop the insurgency. “Simulmatics relied heavily on the work of Pool’s MIT colleague, Lucian Pye, who had argued since the early 1950s that communism was a psychological disease of transitioning peoples. In his influential Politics, Personality, and Nat
ion-Building, he explained that psychological failures lay at the root of stalled nation-building efforts,” writes historian Joy Rohde in “The Last Stand of the Psychocultural Cold Warriors.” To win the war for hearts and minds, Americans needed to design a psychologically appropriate political infrastructure for the emerging nation—a structure through which peasants would develop the appropriate psychological ties to the state.… Military research would write the protocol for a kind of national therapy.”75

  At the same time Simulmatics contractors gathered data in Vietnam’s sweltering jungles, Pool’s company worked on another ARPA initiative called Project ComCom, short for “Communist Communications.” Run out of Pool’s home base at MIT, ComCom was an ambitious attempt to build a computer simulation of the internal communications system of the Soviet Union. The objective was to study the effects that foreign news and radio broadcasts were having on Soviet society as well as to model and predict the kind of reaction a particular broadcast—say, a presidential speech or a breaking news program—would have on the Soviet Union.76 Unsurprisingly, Pool’s models showed that covert CIA attempts to influence the Soviet Union by beaming radio propaganda were having a big effect, and that these efforts needed to be stepped up. “Most of the things of a positive character that are happening in the Soviet Union today are explainable only in terms of the influence of the West, for which the most important single channel is radio,” Pool said in a speech explaining the results of the ComCom study. “In the long run those who are talking to the Soviet Union are not talking to deaf ears. Their voices will be heard and will make a great deal of difference.”77

  But Pool was never satisfied with ComCom’s performance. Even in the late 1960s, the crude state of computer technology meant that it took several months for him and his team to build a model for just one situation.78 It was painstaking work that clearly required more powerful computer tools—tools that simply did not exist.