Surveillance Valley

by Yasha Levine

  Indeed, the army referred to activists and protesters as if they were organized enemy combatants embedded with the indigenous population. They “billeted,” planned assaults on “targets and objectives,” and even had an “organized sniper element.” The army used standard war game colors: blue for “friendly forces” and red for “Negro neighborhoods.” Yet, as the report made very clear, the people being watched were not combatants, but regular people: “Army intelligence was not just reconnoitering cities for bivouac sites, approach routes and Black Panther arsenals. It was collecting, disseminating, and storing amounts of data on the private and personal affairs of law-abiding citizens. Comments about the financial affairs, sex lives, and psychiatric histories of persons unaffiliated with the armed forces appear throughout the various records systems.” That is, the army was spying on a huge swath of American society for no good reason.

  “The hypothesis that revolutionary groups might be behind the civil rights and anti-war movements became a presumption which infected the entire operation,” explained Senator Ervin in a final report his staff produced based on his investigation. “Demonstrators and rioters were not regarded as American citizens with possibly legitimate grievances, but as ‘dissident forces’ deployed against the established order. Given this conception of dissent, it is not surprising that army intelligence would collect information on the political and private lives of the dissenters. The military doctrines governing counterintelligence, counterinsurgency, and civil affairs operations demanded it.”45

  Senator Ervin’s hearings drew a lot of attention and shone a light on the proliferation of federal surveillance databases being assembled in the background unchecked. The army promised to destroy the surveillance files, but the Senate could not obtain definitive proof that the files were ever fully expunged. On the contrary, evidence mounted that the army had deliberately hidden and continued to use the surveillance data it collected.46 Indeed, even as army generals were making promises to destroy files that they had amassed on hundreds of thousands of Americans, ARPA contractors fed them into a new real-time data analysis and retrieval system hooked up to the ARPANET.47

  ARPANET Surveillance

  It was 1975 when NBC aired Ford Rowan’s reporting that the ARPANET was being used to spy on Americans. Three years had passed since Senator Ervin’s investigation of the army’s CONUS Intel spying operation, and the scandal had long become old news, eclipsed by the Watergate investigation that brought down President Richard Nixon. But Rowan’s reporting dragged the sordid CONUS Intel affair back into the spotlight.48

  “In the late 1960’s at the height of the demonstrations against the war, President Johnson ordered the CIA, the FBI, and the Army to find out who was behind the protests. What followed was a major campaign of infiltration and surveillance of antiwar groups,” Rowan told NBC viewers on June 2, 1975. “In 1970, Senator Sam Ervin exposed the extent of Army spying. He got the Pentagon to promise to stop its surveillance program and to destroy the files. But four years after the promise to Sam Ervin, the Army’s domestic surveillance files still exist. NBC News has learned that a new computer technology developed by the Defense Department enabled the Pentagon to copy, distribute, and secretly update the Army files.”

  Two days later, Rowan delivered a follow-up segment:

  The secret computer network was made possible by dramatic breakthroughs in the technique of hooking different makes and models of computers together so they can talk to one another and share information. It’s a whole new technology that not many people know about. If you pay taxes, or use a credit card, if you drive a car, or have ever served in the military, if you’ve ever been arrested, or even investigated by a police agency, if you’ve had major medical expenses or contributed to a national political party, there is information on you somewhere in some computer. Congress has always been afraid that computers, if all linked together, could turn the government into “big brother” with the computers making it dangerously easy to keep tabs on everyone.

  He then got specific about what happened to those surveillance files that the army was supposed to destroy: “According to confidential sources, much of the material that was computerized has been copied and transferred, and much of it has been shared with other agencies where it has been integrated into other intelligence files.… In January 1972, at least part of the computerized Army domestic surveillance files were stored in the NSA’s Harvest computer at Fort Meade, Maryland. Through the use of a defense department computer network, the materials were transmitted and copied in Massachusetts at MIT, and were stored at the Army’s Natick Research Center.”

  The first ARPANET node between UCLA and Stanford went online in 1969 and the network expanded nationally that same year. Now, with Rowan’s exposé six years later, this groundbreaking military network had its first big moment in the public spotlight.

  When I finally tracked down Rowan, he was surprised to hear me bring up that old NBC transmission. No one had discussed it with him in decades. “I haven’t heard anyone who talked about this in a long time. I’m honored that you dug it all out,” he said.

  He then told me how he broke the story.49 In the early 1970s, he was working the Washington beat. He covered Watergate and the Church Committee Hearings run by Senator Frank Church, which remain the most thorough and damning government investigation into the illegal activities of American intelligence agencies, including the CIA, NSA, and FBI. It was during the Church Committee that he first stumbled on the ARPANET story and began piecing it together. “This was post-Watergate, post-Vietnam. This was also the time when they were investigating the assassinations of Kennedy, the assassination of Martin Luther King and later the assassination of Robert Kennedy. Then stories came out about massive domestic spying by the FBI and DoD on antiwar protesters. These investigations were things I was covering and so I would speak to people who were living in that world—the FBI and CIA and the Department of Defense,” Rowan explained. The ARPANET surveillance operation was closely connected to the political upheavals taking place in America at the time, and he learned of its existence in bits and pieces while pursuing other stories. “It was not something that was very easy to find. There was no Deep Throat. No one person that knew it all. You had to really dig.”

  His ARPANET investigation took months to complete. Most sources would not go on the record, but one of them did.50 He was an MIT computer technician named Richard Ferguson, who was there in 1972 when the Pentagon transferred the surveillance data to his lab. He decided to come forward with the information and personally appeared on NBC to make the accusation. He explained that the files were in fact dossiers containing personal information as well as political beliefs. “I’ve seen the data structure that they’ve used and it concerns a person’s occupation, their politics, their name,” he told NBC. He explained that he got fired from his job for objecting to the program.

  Multiple intelligence sources and people involved with the spy file transfer corroborated Ferguson’s claims, but not on the record. In time, other journalists verified Rowan’s reporting.51 There was no doubt: the ARPANET was being used to monitor domestic political activity. “They stressed that the system did not perform any actual surveillance, but rather was designed to use data which had been collected in ‘the real world’ to help build predictive models which might warn when civil disturbances were imminent,” he later wrote in Technospies, a little-known book that expanded on his investigation into the network surveillance technology built by ARPA.52 At least part of the work of writing the database “maintenance program” for the army’s illegal surveillance files appeared to have been carried out at MIT through the Cambridge Project, J. C. R. Licklider’s grand initiative to build computer counterinsurgency data tools.53 They were possibly transferred to other ARPANET sites.

  Harvard and MIT students who protested ARPA’s Cambridge Project back in 1969 saw the ARPANET as a surveillance weapon and a tool of social and political control. They were right. Just a few years
after their protests failed to stop the project, this new technology was turned against them and the American people.

  Ford Rowan’s reporting, and the revelations that the army had not destroyed its illegal surveillance files, triggered another round of congressional investigations. Senator John Tunney, a Democrat from California, led the biggest one. On June 23, 1975, he convened a special session of the Committee on the Judiciary to investigate surveillance technology and to specifically address the role that ARPA’s networking technology played in disseminating the army’s domestic surveillance files.

  Senator Tunney opened the hearings with a condemnation: “We have just gone through a period in American history called Watergate where we saw certain individuals who were prepared to use any kind of information, classified or otherwise, for their own political purposes, in a way that was most detrimental to the interests of the United States and individual citizens,” he said. “We know that the Department of Defense and the Army violated their statutory powers. We know that the CIA violated its statutory power as it related to the collecting of information on private citizens and putting it on computers.”

  He vowed to get to the bottom of the current surveillance scandal to prevent this kind of abuse from happening again and again. For three days, Senator Tunney grilled top defense officials. But just like Senator Sam Ervin, he ran into resistance.54

  Deputy Assistant Secretary of Defense David Cooke, a portly man with a clean-shaven head and slick manner, was one of the main officials representing the Pentagon. He had served under Secretary of Defense Neil McElroy, the man who created ARPA, and he commanded respect and authority. In his testimony, Cooke denied that the army’s domestic surveillance data banks were still in existence, and he doubly denied that the ARPANET had anything to do with transferring or utilizing these nonexistent surveillance files. “Officials at MIT and ARPA state that no transmission of civil disturbance data over ARPANET was ever authorized and they have no evidence it ever occurred,” he testified. He also did his best to convince Senator Tunney that the Pentagon had no operational need for the ARPANET, which he described as a pure research and academic network. “The ARPANET itself is a totally unclassified system, which was developed by and is widely utilized by the scientific and technological community throughout the United States,” he told the committee. “Neither the White House nor any of the intelligence agencies has a computer connected to the ARPANET.”

  As Cooke explained it, the military did not need the ARPANET because it already had its own secure database and network for communication and intelligence files: the Community Online Intelligence System, known simply as COINS. “It is a secure system, connecting selected data banks of three intelligence agencies, the Defense Intelligence Agency, National Security Agency, and the National Photo Interpretation Center. It is designed to exchange classified and highly sensitive foreign intelligence data among these intelligence agencies and within the Department of Defense. The Central Intelligence Agency and Department of State can access the system,” he explained, and then added emphatically: “COINS and the ARPANET are not linked, and will not be linked.”

  He was either misinformed or stretching the truth.

  Four years earlier, in 1971, ARPA director Stephen Lukasik, who had run the agency during the build-out of the ARPANET, very clearly explained in his testimony to the Senate that the whole point of the ARPANET was to integrate government networks—both classified (like COINS) and unclassified—into a unified telecommunications system.55 “Our objective is to design, build, test and evaluate a high-performance, low-cost, reliable computer network to meet the growing DoD requirements for computer-to-computer communications,” he said. He added that the military had just started testing the ARPANET as a way to connect operational computer systems.56

  Acording to Lukasik, the beauty of the ARPANET was that, although it was technically an unclassified network, it could be used for classified purposes because data could be digitally encrypted and sent over the wire without the need to physically secure actual lines and equipment. It was a general-purpose computer network that could connect to public networks and be used for classified and nonclassified tasks.57

  Lukasik was right. Between 1972 and 1975, multiple military and intelligence agencies not only connected directly to the ARPANET but also began to build their own operational subnetworks that were based on the ARPANET design and that could interconnect with it. The navy had multiple air bases tied to the network. The army used the ARPANET to link supercomputer centers. In 1972, the NSA had commissioned Bolt, Beranek and Newman—J. C. R. Licklider’s company and major ARPANET contractor—to build an upgraded ARPANET version of its COINS intelligence system, the very system that Cooke promised three years later would never be plugged into the ARPANET. This system ended up being connected to the ARPANET to provide operational data communication services for the NSA and the Pentagon for many years afterward.58

  Even as Cooke denied to Senator Tunney that the ARPANET was used for military communications, the network featured multiple connections from the army, navy, NSA, and air force—and very likely contained unlisted nodes maintained by intelligence agencies such as the CIA.59 But the issue soon became moot. A few weeks after Cooke’s testimony, the ARPANET was officially absorbed by the Defense Communications Agency, which ran the communications systems for the entire Pentagon. In other words, even if still somewhat experimental, the ARPANET was the definition of an operational military network.60

  Military Internet

  In the summer of 1973, Robert Kahn and Vint Cerf locked themselves in a conference room at the upscale Hyatt Cabana El Camino Real just a mile south of Stanford. The Cabana was the most glamorous hotel in Palo Alto, having hosted the Beatles in 1965, among other celebrities. Kahn was stocky and had thick black hair and sideburns. Cerf was tall and lanky, with an unkempt beard. The two could have been a folk music duo passing through on tour. But Kahn and Cerf weren’t there to play or socialize or party. They didn’t have any booze or drugs. They didn’t have much more than a few pencils and pads of paper. For the past several months, they had been trying to create a protocol that could connect three different types of experimental military networks. At the Cabana, their mission was to finally get their thoughts on paper and hash out the final technical design of an “inter-net.”61

  “Do you want to start or shall I?” asked Kahn.

  “No, I’ll be happy to begin,” Cerf replied, and then sat there staring at a blank piece of paper. After about five minutes, he gave up: “I don’t know where to start.”62

  Kahn took over and scribbled away, jotting out thirty pages of diagrams and theoretical network designs. Both Cerf and Kahn had been involved in building the ARPANET: Cerf had been part of a UCLA team responsible for writing the operating system for the routers that formed ARPANET’s backbone, while Kahn had worked at Bolt, Beranek and Newman helping to design the network’s routing protocols. Now they were about to take it to a new level: ARPANET 2.0, a network of networks, the architecture of what we now call the “Internet.”

  In 1972, after Kahn was hired to head ARPA’s command and control division, he had convinced Cerf to leave a job he had just taken teaching at Stanford and work for ARPA again.63 A major goal for Kahn was to expand the ARPANET’s usefulness in real-world military situations. That meant, first and foremost, extending the packet-based networking design to wireless data networks, radio, and satellite. Wireless data networks were crucial to the future of military command and control because they would allow traffic to be transmitted over huge distances: naval vessels, aircraft, and mobile field units could all connect to computer resources on the mainland through portable wireless units. It was a mandatory component of the global command and control system ARPA was charged with developing.64

  Kahn directed the effort to build several experimental wireless networks. One was called PRNET, short for “packet radio network.” It had the ability to transmit data via mobile computers installed in vans
using a network of antennas located in the mountain ranges around San Bruno, Berkeley, San Jose, and Palo Alto. The effort was run out of the Stanford Research Institute. At the same time, Kahn pushed into packet satellite networking, setting up an experimental network called SATNET that linked Maryland, West Virginia, England, and Norway; the system was initially designed to carry seismic data from remote installations set up to detect Soviet nuclear tests. ARPANET’s data packet technology worked remarkably well in a wireless setting. But there was one problem: although they were based on the same fundamental data-packet-switching designs, PRNET, SATNET, and ARPANET all used slightly different protocols to run and so could not connect to each other. For all practical purposes, they were standalone networks, which went against the whole concept of networking and minimized their usefulness to the military.

  ARPA needed all three networks to function as one.65 The question was: How to bring them all together in a simple way? That’s what Kahn and Cerf were trying to figure out in the Cabana conference room. Eventually, they settled on a basic plan for a flexible networking language that could connect multiple types of networks. It was called TCP/IP—Transmission Control Protocol/Internet Protocol, the same basic network language that powers the Internet today.66

  In a 1990 oral history interview, Cerf, who now works as Google’s Chief Evangelist, described how his and Kahn’s efforts to devise an internetwork protocol were entirely rooted in the needs of the military:

  There were lots of ramifications for the military. For example, we absolutely wanted to bring data communications to the field, which is what the packet radio project and the packet satellite projects were about; how to reach wide areas, how to reach people on the oceans. Can’t do it by dragging fiber, can’t do it very well with terrestrial store-and-forward radio because line-of-site doesn’t work very well on a wide ocean. So you need satellites for that. So the whole effort was very strongly motivated by bringing computers into the field in the military and then making it possible for them to communicate with each other in the field and to assets that were in the rear of the theatre of operations. So all of the demonstrations that we did had military counterparts.67