INFORMATION SCIENCES : Thirty Years Ago, a Handful of Professors and Grad Students Cobbled Together a Clunky Computer Network. Today, Some of These Internet Kings Dream of Wiring the Solar System.
The machine arrived air express from Boston the Saturday before Labor Day, 1969. It was a great steel behemoth, with an armored exterior and four steel hooks on top, which was lucky, because it wouldn’t fit in the elevator of UCLA’s Boelter Hall, and its handlers feared they might have to winch it up the side of the building by crane. Instead they managed to get it up to an interior catwalk on a three-story forklift and into Room 3400.
“We didn’t even have a camera,” Leonard Kleinrock recalls. “Edison would have. Samuel Morse, too, probably. But we didn’t.”
He speaks with palpable resignation about this opportunity lost forever to mark the birth of a new age. The lightbulb, the telegraph . . . the Web. The machine being installed that summer day in UCLA’s engineering building, under the eyes of a handful of professors and graduate students, was Interface Message Processor #1--the first of the engagingly named IMPs to be installed anywhere. These were the machines that formed the backbone of what was then known as the ARPANET. The name came from ARPA, the Defense Department’s Advanced Research Projects Agency, and was to morph, many years later, into the Internet.
Kleinrock was in charge of the team of grad students writing the software for the first IMP and overseeing its installation. Then, as now, he was a bantamweight dynamo, compact and wiry (he’s a black belt in karate and a marathon runner), with a Cagneyesque delivery that still betrays his New York origins. He is also a brilliant theoretical engineer whose 1963 MIT doctoral dissertation laid out a system of transmitting digital data over networks that would remain the standard for decades.
To Len Kleinrock, there can be no argument that Labor Day, 1969, marks the Internet’s birth and, therefore, validates UCLA’s claim to be its cradle.
Not everyone buys that claim. “The Internet has had a lot of midwives and it still continues to be reborn,” says Vinton G. Cerf, an executive at MCI WorldCom who was present that Labor Day as one of Kleinrock’s graduate students.
Suffice to say that something extraordinary was born that weekend in Southern California, even if none of those in attendance claim to have had the foresight to imagine the new medium’s commercial and informational reach.
The team had come together by chance and opportunity, many of them attracted by UCLA engineering professor Gerald Estrin. There were Vint Cerf and Stephen D. Crocker, best friends from Van Nuys, and Jonathan Postel--another graduate of Van Nuys High School, a shy man with a rabbinical beard who would grow up to become one of the Internet’s worldwide gatekeepers.
In 1969 they were very much guides without a road map. No one had constructed a large-scale computer network before. Nor was the network the focus of a great deal of technical interest.
“Among grad students, the hot topics were programming languages, artificial intelligence, graphics and machine architecture,” Crocker says. “Networking didn’t have the sizzle for a majority of people. I suffered from similar thoughts--I used to say the problem with it was it only had socially redeeming features.”
If the grad students were indifferent, their professors were outright hostile. ARPA’s network was a project they feared would drain their precious research budgets or, worse, allow outside institutions to poach their computer resources. This attitude was especially prevalent among Eastern universities like MIT, where the computer centers were older, established and better-funded.
That doesn’t mean that West Coast computer faculties were entirely free of opposition to sharing these multimillion-dollar treasures. That had become clear in 1965, when ARPA offered to finance a project to link three identical IBM 704s belonging to UCLA--at the engineering department, the business school and the campus computer center. The plans led to an outbreak of intramural jealousies and disagreements so severe that, to ARPA’s astonishment, the university turned down its offer--reportedly $2 million. “It was sheer politics,” Kleinrock says. “We never even got to the technical issues.”
Therefore, it is no surprise that the idea of sharing computing cycles with alien institutions was an unpopular one. Nevertheless, ARPA’s force of will (and its money) soon prevailed. Four West Coast institutions--UCLA, the private Stanford Research Institute in Menlo Park, UC Santa Barbara and the University of Utah--were signed up as the first four nodes, with UCLA winning responsibility to monitor and test the overall network’s performance. Characteristically, the professors who were the agency’s formal grant recipients delegated the design scut work to their students, who began mapping out what would become a world-altering new communications medium in a series of informal get-togethers.
“There was no coherence to the meetings, not even an organized set of questions,” recalls Crocker. On the other hand, they discovered a like-mindedness that would ease the difficult technical task ahead.
By mid-1969, Kleinrock’s team was hard at work on questions much tougher than those that might have arisen from linking together three identical UCLA computers. The ARPANET would require the sharing of information by scores of architecturally distinct, and therefore incompatible and mutually unintelligible computers, ranging from IBM mainframes to the Scientific Data Systems Sigma-7 at UCLA.
The IMPs were the solution: Reconfigured Honeywell minicomputers, they were to function like interpreters at the United Nations, translating each host’s gabble into a common network Esperanto for transmission to other IMPs and subsequent retranslation for their own hosts. But no one at UCLA could know if the IMP would actually succeed in talking to the Sigma-7, much less get an answer, until it was tried.
“It was the day after Labor Day, Tuesday, Sept. 2, and a lot of people were here,” Kleinrock says from his Boelter Hall office, a short amble down the corridor from the room where IMP #1 still stands as a museum piece. “We were going to turn it on and try to move bits from the IMP to the host and back. And I mean just bits, I don’t even mean messages that meant anything. We had the people from ARPA, from Honeywell, people from AT&T; and GTE [the system was to run on phone lines], from UCLA. We were all worried that it wouldn’t work. But bits flowed that first day. The next day, messages flowed between the two machines--things that at least made some sense.”
One would have to note here that this was not yet exactly on a global scale. The IMP and its host were standing in the same room, 20 feet apart. But a month later a second IMP was installed at the Stanford Research Institute, and a month after that IMP #3 at UC Santa Barbara. The network was for real, allowing researchers at those sites to use each other’s computers as if they were all one. By mid-1970, there were 10 operating nodes, with the East Coast and West Coast sites connected by two high-performance cross-country phone lines.
The number of useful applications grew quickly--ways to transfer files and e-mail, for example. But the vast potential of their offspring almost always progressed faster than their own vision. “We were able to see pretty far ahead, but not forever,” Crocker says. “It was no mystery that there’d be a nationwide network connecting research centers, but e-mail was nowhere on our radar screens.”
Indeed, the addition of e-mail was one watershed that demonstrated how useful the net might prove. “I was less conscious of the communications capability until then--but then there was no doubt in anyone’s mind that this was very powerful stuff,” Cerf says. Still, he adds, “it took me until 1988 to believe that this could be of commercial interest, as opposed to a marvelous research tool.”
Meanwhile the ARPANET’s original midwives were moving on, becoming more aware almost daily of the network’s growing power. In 1973, with a fellow engineer named Bob Kahn, Cerf developed the so-called “transmission control protocol,” or TCP, which allows data to flow reliably across networks of almost any size. That was a crucial step in transforming the ARPANET into the Internet (Cerf himself regards it as a pretty fair candidate for the birth date of the Internet). It’s still in use today.
In time, Cerf and Crocker both would join ARPA (by then renamed the Defense Advanced Research Projects Agency, or DARPA) to help raise the newborn net to adolescence. Then they moved on again, Crocker to the nonprofit Aerospace Corporation as director of its computer science laboratory and Cerf to MCI, where he was chief engineer for MCI Mail, the first private application to utilize the Internet. Meanwhile Kleinrock co-founded a company called Linkabit, and later another called Nomadix, which develops technology enabling people to remain connected to their networks no matter where they roam.
Their UCLA colleague Jon Postel joined USC’s Information Sciences Institute. From there he served as caretaker and implementer of the Internet’s technical standards--the Internet’s chief plumber, as one fellow pioneer put it. It was a job that often required him to defend the architectural principles on which the network was founded from the intensifying incursions of commercial and political interests--something that often placed the shy Van Nuys native at the center of multiple controversies.
“He was quiet, patient, and well, stubborn, too,” Bob Metcalfe, a networking pioneer who invented the Ethernet in 1973 wrote not long after Postel’s death last October. “But the only difference between being stubborn and being a visionary is whether you are right--and he was, much more often than not.”
Those who survived him retain a forefathers’ influence over the network they nurtured. Take Vint Cerf: Popular as a speaker on the technology-and-science conference circuit, he began mentioning in late 1997 how the Internet’s communications capabilities might serve mankind in its exploration of the solar system. (A science-fiction enthusiast, Cerf consulted with the late Gene Roddenberry on the creation of the TV show “Earth: Final Conflict.”)
As it happened, NASA was at the time considering the same issues. The agency was trying to communications systems of all the exploratory satellites were incompatible. Cerf’s thinking promptly came to the attention of Adrian J. Hooke, the man in charge of standardizing spacecraft communications systems at NASA’s Jet Propulsion Laboratory in Pasadena.
“It was quite eerie when we realized the Internet guys were working on the same things as we were,” says Hooke.
Today Hooke, the planetary expert, and Cerf, the Internet visionary, are working together on what is likely to be a future generation of the Internet--in both its terrestrial and cosmic incarnations. Both understand that each side is certain to gain from shared development of the so-called Interplanetary Internet.
“Space is one of those honing environments,” Hooke says from his office at JPL. “There’s a real symbiosis here.” More specifically, he observes, an Internet spanning the distance from Earth to, say, robot stations or even the manned stations NASA hopes to establish on Mars, will face the same technical problems confronted by a terrestrial Internet trying to accommodate an exploding quantity of data passing through satellite links to laptops and mobile phones. “We’ve been dealing with characteristics for 30 years that these ‘edge’ markets are just now facing.”
From there the vista expands exponentially. Where at first the network’s founders were hard pressed to see where it would be two or three years ahead, today the future seems almost limitless. The vision of people always on the move yet always within reach--of their loved ones, their work, their entertainment--is merely the start.
