Tubes: A Journey to the Center of the Internet

by Andrew Blum

  2

  * * *

  A Network of Networks

  I wanted to know where the Internet started, but the question turned out to be more complicated than I’d imagined. For an invention that dominates our daily lives—acknowledged as an epoch-making transformative force across global society—the Internet’s history is surprisingly underwritten.

  The serious book-length histories all seem to have been published in 1999, as if the Internet were finished then—as if the Internet were finished now. But more than that timing, they each seem to have their own heroes, milestones, and beginnings. The Internet’s history, like the network, was itself distributed. As one historian of historians put it (writing in 1998), “the Internet lacks a central founding figure—a Thomas Edison or a Samuel F. B. Morse.” I should have known things wouldn’t be so clear-cut when the author of Inventing the Internet, widely considered the most authoritative among them, began by suggesting that “the history of the Internet holds a number of surprises and confounds some common assumptions.” I felt like a guy wandering around a party he wasn’t invited to, asking who the host was, and nobody knew. Or maybe there wasn’t a host at all? Maybe the problem was more philosophical than that? The Internet had a chicken-and-egg thing going on: If the Internet is a network of networks, then it takes two networks to make an Internet, so how could one have been the first?

  Needless to say, all this did not inspire confidence. I had set out in search of the real, the concrete, the verifiable, but I was greeted at the door by the historiographic equivalent of a comments thread. My question had to be narrower, more rooted in time and place. It was about the object. “Not ideas about the thing but the thing itself,” as Wallace Stevens wrote. Not, where did the Internet begin? But, where was its first box? And that, at least, was clear.

  In the summer of 1969, a machine called an interface message processor, or IMP, was installed at the University of California–Los Angeles, under the supervision of a young professor named Leonard Kleinrock. He’s still there, a little less young, but with a boyish smile and a website that seemed to encourage visitors. “You’ll want to meet me in my office,” he replied when I emailed. “The original site of the IMP is just down the hall.” We made arrangements. But it wasn’t until I settled into my cramped seat on the plane to Los Angeles, surrounded by tired consultants in wrinkled shirts and aspiring starlets in sunglasses, that the full implications of my journey sank in: I was going to visit the Internet, flying three thousand miles on a pilgrimage to a half-imagined place. And what did I expect to find? What, truly, was I looking for?

  I suppose all pilgrims feel that way at some point. We are optimistic creatures. In Judaism, the Temple Mount in Jerusalem is the place from which the whole world expanded, the place closest to God, and the most important place of prayer. For Muslims, the small cube building in Mecca known as the Kaaba is the holiest place, so dominant in the psychic geography of the devout that they face it to pray five times a day, wherever they are in the world, even flying across the ocean on an airplane. Every cult, group, team, gang, society, guild—whatever—has its significant place, marked with memory and meaning. And most of us also have our own individual places: a hometown, stadium, church, beach, or mountain that looms epically above our lives.

  Yet this significance is always in a way personal, even if millions collectively share it. Philosophers like to point out that “place” is as much within us as without us. You can demarcate a place on a map, pinpoint its latitude and longitude with global positioning satellites, and kick the very real dirt of its very real ground. But that’s inevitably going to be only half its story. The other half of the story comes from us, from the stories we tell about a place and our experience of it. As the philosopher Edward Casey writes, “Stripping away cultural or linguistic accretions, we shall never find a pure place lying underneath.” All we shall find instead are “continuous and changing qualifications of particular places.” When we travel, we fix a place’s meaning in our minds. It is in the eyes of a pilgrim that a holy site becomes holiest. And in being there, he affirms not only the place’s significance but also his own. Our physical place helps us better know our psychic place—our identity. But did that hold true for me, on the path to the Internet? I longed to see its most significant places, but were its places really places at all? And if they were, was the Internet close enough to religion—a way of understanding the world—that seeing those places would be meaningful?

  The question got more complicated the next morning in Los Angeles. I woke up at dawn, my body clock still set to New York, in an enormous hotel near the airport, with a mirrored glass façade and a view of the runway. I stood in front of the window watching a line of jets land on top of their shadows. Inside, nearly every surface had a little folded cardboard sign that indicated the branded touches that made the room conform to the hotel chain’s international standards: the “Suite Dreams®” bed, the “Serenity Bath Collection™,” the “Signature Service.” Nothing was singular or local; everything came from far away, at the direction of a global corporation. The novelist Walter Kirn calls this “Airworld”—these nonplaces of airports and their surroundings. I tried to get a little postmodern kick out of it all and summon my inner Ryan Bingham, the protagonist of Kirn’s novel Up in the Air (played by George Clooney in the movie) who only feels at home here in this homogeneous, if admittedly comfortable, world—even if “cities don’t stick in my head the way they used to.” But I was hollow. On my way to the Internet I was already climbing a steep slope toward the singular and the local. It was frustrating to find the ostensibly real places blurring into each other as well. I had come to Los Angeles to try to bring the network back into the world—but instead, the world seemed to have succumbed to the logic of the network.

  But I shouldn’t have worried. At UCLA that afternoon, the moment of the Internet’s physical birth came vividly into focus, rooted in a very specific place. On the quiet Saturday afternoon of the Labor Day weekend in 1969, a small crowd of computer science graduate students had gathered in the courtyard of Boelter Hall with a bottle of champagne. Standing in the same spot, I conjured the scene. The occasion was the arrival of their grand and expensive new gadget, coming that day from Boston by air freight: a modified and military-hardened version of a Honeywell DDP-516 minicomputer—“mini,” at the time, meaning a machine that weighed nine hundred pounds and cost $80,000, the equivalent of nearly $500,000 today. It was traveling from the Cambridge, Massachusetts, engineering firm of Bolt, Beranek and Newman, possessor of a $1 million Department of Defense contract to build an experimental computer network, known as the ARPANET. Among Bolt’s many customizations was a new name for the machine: the Interface Message Processor. The particular one then wending its way up the hill to the UCLA campus was the very first: IMP #1.

  The graduate students were mostly my parents’ age, born in the last days of World War II—proto–baby boomers, in their midtwenties at the time—and I can blur in some family photos of the era. It was the summer of Woodstock and men on the moon, and even computer scientists wore their hair shaggy and their pants legs wide. One probably had a button with the word RESIST on it next to a question mark, the scientific notation for electrical resistance and a popular antiwar symbol for engineers. They all knew that their funding, $200,000 to UCLA alone supporting forty grad students and staff, came from the Department of Defense. But they also all knew that what they were building wasn’t a weapon.

  The ARPANET project was managed by the Department of Defense’s Advanced Research Projects Agency (ARPA), founded in the wake of Sputnik’s launch to support scientific research, esoteric stuff far out on the technological frontier. ARPANET certainly qualified. There had been few attempts to connect computers at continental-scaled distances, let alone create an interconnected network of them. If somewhere deep in the Pentagon a four-star general had the grim notion that the nascent ARPANET might evolve into a communications network that could survive a nuclear wa
r—a popular myth about the origins of the Internet—this group was insulated from it. And anyway, they ignored it. They were consumed by the technical challenges arriving inside that moving van, by new wives and babies, and by the infinite possibilities of computer communications. Consumed, that is, by peaceful intentions.

  Boelter Hall was new and shiny then, like much of Los Angeles itself. Built in the early 1960s to house the rapidly expanding engineering department, its stripped modernist lines were the height of architectural fashion, suitable to the cutting-edge work going on inside—not unlike the new biomolecular science building that now towers over it next door. These days Boelter is a bit rough around the edges, with worn-out sunshades over the windows and rusting steel balcony railings facing a courtyard of full-grown eucalyptus trees. The IMP’s welcoming party would have stood here beneath them in the shade, on a hot Southern California day. Long before cell phones, they would have guessed at the timing of the truck’s progress from the airport. A forklift waited nearby, ready to lift the massive machine up into the building. Did they sip their champagne from Styrofoam cups? Snap pictures with one of the inexpensive new Japanese cameras that had recently begun to be imported? (If so, they’re long lost.) The excitement of the occasion would have been unmistakable, even if the full historic implications were not: this was the first piece of the Internet.

  But while the grad students were celebrating outside, their professor was stuck upstairs, alone in the large office he had recently expanded in a fit of empire building, shuffling papers on a Saturday afternoon. This I can picture precisely, because when I walked in forty-one years later, Leonard Kleinrock was still sitting there, sprightly at seventy-five, wearing a starched pink shirt, black slacks, and a BlackBerry clipped to a polished leather belt. His face was tanned and his hair was full. A brand-new laptop was open on his desk and he was yelling into a speakerphone: “It’s not catching!”

  On the other end, the disembodied voice of a tech support person responded slowly and patiently. Click here. Now click there. Type this in. Kleinrock looked over the top of his reading glasses and waved me toward a chair. Then he clicked. And clicked again.

  Now try, the voice said.

  He winced. “It says I’m not connected to the Internet. That’s what it says!” Then he laughed so hard his shoulders shook.

  Kleinrock is the father of the Internet—or rather, a father, as success has many. In 1961, while a graduate student at MIT, he published the first paper on “packet switching,” the idea that data could be transmitted efficiently in small chunks rather than a continuous stream—one of the key notions behind the Internet. The idea was already in the air. A professor at the British National Physical Laboratory named Donald Davies had, unbeknownst to Kleinrock, been independently refining similar concepts, as had Paul Baran, a researcher at the RAND Corporation in Los Angeles. Baran’s work, begun in 1960 at the request of the US Air Force, was explicitly aimed at designing a network that could survive a nuclear attack. Davies, working in an academic setting, merely wanted to improve England’s communications system. By the mid-1960s—by which time Kleinrock was at UCLA, on his way toward tenure—their ideas were circulating among the small global community of computer scientists, hashed out at conferences and on office chalkboards. But they were only ideas. No one had yet fit the pieces of the puzzle together into a working network. The fundamental challenge these network pioneers faced—and the one that remains at the heart of the Internet’s DNA—was designing not just a network but a network of networks. They weren’t only trying to get two or three or even a thousand computers talking, but two or three or a thousand different kinds of computers, grouped in all sort of ways, spread far and wide. This metalevel challenge was known as “Internetworking.”

  It took the Department of Defense to bring teeth to it. In 1967, a young computer scientist named Larry Roberts—Kleinrock’s MIT office mate—was recruited to ARPA specifically to develop an experimental nationwide computer network. The next July, he sent out a detailed request for proposals to 140 different technology companies to build what he at first called the “ARPA net.” It would begin at four universities, all in the west: UCLA, Stanford Research Institute, the University of Utah, and the University of California–Santa Barbara. The geographic bias wasn’t an accident. Connecting university computers together was a threatening idea—each school would inherently have to share its prized, and already overused, machine. The East Coast schools tended to be more conservative, or at least less susceptible to Roberts’s ability to influence them with his control over their ARPA funding. California already had a burgeoning technology culture and big universities, but the nascent ARPANET’s West Coast beginnings had as much to do with a cultural appetite for new ideas.

  Under Kleinrock’s direction, UCLA would have the added task of hosting the Network Measurement Center, responsible for studying the performance of this new creation. The reason for that was as much personal as professional: not only was Kleinrock the reigning expert in network theory but Roberts trusted his old friend. If Bolt, Beranek and Newman’s job was to build the network, Kleinrock’s would be to break it, to test the limits of its performance. It also meant that UCLA would receive that first IMP, to be installed between the computer science department’s big, shared computer, called a Sigma-7, and the specially modified phone lines to the other universities that AT&T had readied for when there eventually was a network. For its first month in California, IMP #1 stood alone in the world, an island awaiting its first link.

  “Do you want to see it?” Kleinrock asked, jumping excitedly from his chair. He led me across the hall to a small conference room, not fifteen feet from his desk. “This is it—a beautiful machine! A really magnificent machine!” The IMP looked—like all famous things—exactly like it does in photographs: refrigerator-sized, beige, steel, with buttons on the front, like a file cabinet dressed up as R2-D2. He opened and closed the cabinet and twisted some dials. “Military hardened, built out of a Honeywell DDP-516, state of the art at the time.” I said I had begun to notice that the Internet had a smell, an odd but distinctive mix of industrial-strength air conditioners and the ozone released by capacitors, and we both leaned in for a whiff. The IMP smelled like my grandfather’s basement. “That’s mildew,” Kleinrock said. “We should close the door and it’ll cook it up.”

  The IMP’s current situation certainly lacked in ceremony, shoved as it was into the corner of a small conference room, with mismatched chairs and faded posters on the walls. A stack of paper coffee cups poked out of a plastic bag. “Why is it sitting here?” Kleinrock said. “Why is it not in a wonderful showcase somewhere on campus? The reason is nobody took this machine as being important. They were going to throw it out. I had to rescue it. Nobody recognized its value. I said, ‘We gotta keep this thing, it’s important!’ But you’re never a philosopher in your own country.”

  But that was changing. A history graduate student at UCLA had recently clued into the historical significance of Boelter Hall and the IMP and had begun assembling archival materials. After years of pleading with the university administration, Kleinrock had finally gathered support for the construction of the Kleinrock Internet Heritage Site and Archive. It would commemorate not only the IMP itself but the historical moment. “It was amazing, this group of really smart people collected in the same time and the same place,” Kleinrock said. “It happens, it’s sort of periodic, when you get this kind of golden era.” Indeed, the group assembled in his lab that fall formed a core group of Internet hall of famers, notably Vint Cerf (now “Chief Internet Evangelist” at Google), who cowrote the Internet’s most important operational code—what is known as the TCP/IP protocol—with Steve Crocker, also Kleinrock’s student, and Jon Postel, who managed the Internet Assigned Numbers Authority for years and was a key mentor to an entire generation of network engineers.

  The museum would be built in room 3420, where the IMP had been installed from Labor Day 1969 until it was decommissioned in
1982. We walked down the hall to see it. “The IMP was against this wall here,” Kleinrock said, slapping the white paint, “but the room has been reconfigured. The ceiling is new, the floor is new—we had a raised floor for air-conditioning.” We peered behind a steel storage cabinet to see if the original phone jack might still be there—the first few feet of the Internet’s first route—but it wasn’t. There was no plaque, no historic display, and there were certainly no tourists, not yet. Kleinrock’s hope was to restore the room to the way it looked in 1969, which I imagined would become something like Graceland, frozen in time, with the IMP and an old rotary telephone and photographs of men in heavy-framed glasses and with slicked-down hair. “To put a wall up here and put a doorway there is $40,000, and we have a $50,000 budget for this and the archivist,” Kleinrock said. “So I’m going to have to donate a lot of money, I think. It’s okay. It’s a good cause.”

  As we talked, an undergraduate computer lab was in progress in the room, with students touching soldering irons to green circuit boards, their cell phones set on the desks in front of them, while a teaching assistant barked instructions. No one even glanced at us. Kleinrock was one of the Internet’s earliest masterminds, but to the nineteen-year-olds in here, whose lives were fully shaped by its presence—Internet Explorer came out before they would have learned to read—he faded into the woodwork. Almost literally. This wasn’t a holy site, it was a classroom—far less of a tourist attraction than Ryan Seacrest’s house, not far away. So what was I doing here?