Pioneers of the Internet

The Internet is unique among human creations in many ways. There’s only one of it. It had many inventors, rather than just one. We perceive it more as a community than as a thing. But unlike most communities, it charges no admission fee, collects no taxes, imposes no rules, and has no visible authority. You can use it to learn, to laugh, to chat, to buy; or you can use it to spam, to flame, to steal, to spy. The Internet doesn’t care. It exists only to pass messages, not to pass judgment.

What isn’t so widely known is that the Internet acquired its free-for-all nature by design. Its founders wanted no barriers to access, with minimal intervention by software in the middle and maximal autonomy for the users at the end. Two of these anonymous architects, David Clark ’66 and Stephen Wolff ’57, were named to the Internet Hall of Fame in 2013. At a meeting in 1992, Clark summed up the Internet’s governing philosophy: “We reject: kings, presidents, and voting. We believe in: rough consensus and running code.” It’s a remarkable statement, combining the Quaker philosophy of consensus with the engineering philosophy that whatever works, wins. 

Wolff and Clark started work on the Internet before there was an Internet. Wolff began at the Army Ballistic Research Laboratories (BRL) in Aberdeen, Md. As one of 30 hubs on the ARPANET, the precursor to the Internet, Aberdeen was supposed to upgrade its software on New Year’s Eve 1980—but with a week to go, they weren’t ready. The problem was thrown into Wolff’s lap, and he asked computer scientist Mike Muuss to help. Muuss declared the existing software to be “junk,” and reprogrammed the whole hub in a marathon session. “We made the changeover an hour before midnight,” Wolff says. “Mike called it the BRL Unix Networking System, or BUNS for short.” When he was done, he emailed Wolff’s supervisor: “BRL put its BUNS online.”

The anecdote says a lot about the early days of networking. Computers were not designed to communicate with one another. The software to do so had to be put together from scratch. Vint Cerf of Stanford and Robert Kahn of the Defense Advanced Research Projects Agency (DARPA) had a vision for linking them, which they called the Transmission Control Protocol and the Internet Protocol (TCP/IP). They published the first specifications for TCP/IP in 1974, but it was a long way from a working Internet.

While Wolff was saving BRL’s buns, Clark was a postdoc at MIT, turning Cerf and Kahn’s vision into a reality. “Several of us were writing the Internet Protocol,” he says. Each version ran on a different brand of computer—one for IBMs, one for PDPs, and so on. It took six years to get all the computers working together. 

Then, in 1981, with TCP/IP finally up and running, the Internet reached its first great transition: Its leader left. “Vint was the intellectual leader and was also in charge of the money, who corralled us both by force of his character and by threat of defunding,” Clark says. No one person could replace Cerf, but for a while two people did: Clark as the visionary and Wolff (a few years later) as the source of funds.

Cerf anointed Clark as the chief protocol architect of the Internet, a title he held from 1981 to 1989 and no one has held since. His successors are called chairs of the Internet Architecture Board (IAB). Clark imparted the Internet’s ethos and design principles to newcomers. When engineers came to the IAB with new ideas, he and the other board members could usually tell what would work and what wouldn’t.

Meanwhile, in 1985, Wolff arrived at the National Science Foundation (NSF) on the cusp of three great changes. 

The Internet is unique among human creations in many ways. There’s only one of it. It had many inventors, rather than just one. We perceive it more as a community than as a thing. But unlike most communities, it charges no admission fee, collects no taxes, imposes no rules, and has no visible authority. You can use it to learn, to laugh, to chat, to buy; or you can use it to spam, to flame, to steal, to spy. The Internet doesn’t care. It exists only to pass messages, not to pass judgment.

What isn’t so widely known is that the Internet acquired its free-for-all nature by design. Its founders wanted no barriers to access, with minimal intervention by software in the middle and maximal autonomy for the users at the end. Two of these anonymous architects, David Clark ’66 and Stephen Wolff ’57, were named to the Internet Hall of Fame in 2013. At a meeting in 1992, Clark summed up the Internet’s governing philosophy: “We reject: kings, presidents, and voting. We believe in: rough consensus and running code.” It’s a remarkable statement, combining the Quaker philosophy of consensus with the engineering philosophy that whatever works, wins. 

Wolff and Clark started work on the Internet before there was an Internet. Wolff began at the Army Ballistic Research Laboratories (BRL) in Aberdeen, Md. As one of 30 hubs on the ARPANET, the precursor to the Internet, Aberdeen was supposed to upgrade its software on New Year’s Eve 1980—but with a week to go, they weren’t ready. The problem was thrown into Wolff’s lap, and he asked computer scientist Mike Muuss to help. Muuss declared the existing software to be “junk,” and reprogrammed the whole hub in a marathon session. “We made the changeover an hour before midnight,” Wolff says. “Mike called it the BRL Unix Networking System, or BUNS for short.” When he was done, he emailed Wolff’s supervisor: “BRL put its BUNS online.”

The anecdote says a lot about the early days of networking. Computers were not designed to communicate with one another. The software to do so had to be put together from scratch. Vint Cerf of Stanford and Robert Kahn of the Defense Advanced Research Projects Agency (DARPA) had a vision for linking them, which they called the Transmission Control Protocol and the Internet Protocol (TCP/IP). They published the first specifications for TCP/IP in 1974, but it was a long way from a working Internet.

While Wolff was saving BRL’s buns, Clark was a postdoc at MIT, turning Cerf and Kahn’s vision into a reality. “Several of us were writing the Internet Protocol,” he says. Each version ran on a different brand of computer—one for IBMs, one for PDPs, and so on. It took six years to get all the computers working together. 

Then, in 1981, with TCP/IP finally up and running, the Internet reached its first great transition: Its leader left. “Vint was the intellectual leader and was also in charge of the money, who corralled us both by force of his character and by threat of defunding,” Clark says. No one person could replace Cerf, but for a while two people did: Clark as the visionary and Wolff (a few years later) as the source of funds.

Cerf anointed Clark as the chief protocol architect of the Internet, a title he held from 1981 to 1989 and no one has held since. His successors are called chairs of the Internet Architecture Board (IAB). Clark imparted the Internet’s ethos and design principles to newcomers. When engineers came to the IAB with new ideas, he and the other board members could usually tell what would work and what wouldn’t.

Meanwhile, in 1985, Wolff arrived at the National Science Foundation (NSF) on the cusp of three great changes. DARPA was about to end its support of the ARPANET. At the same time, academic researchers were eager to get connected. Finally, the NSF wanted to hook its five brand-new supercomputer centers to the ARPANET backbone. Instead, the high-speed links connecting San Diego, Urbana, Cornell, Pittsburgh and Princeton universities became the new backbone of the NSFNet. This network bridged the gap from the Internet’s infancy to its maturity.

As program manager at NSF, Wolff coaxed the Internet out of its academic shell. Although the NSF could fund the backbone, it couldn’t fund all the regional networks. “I told them that they should go out and get commercial customers,” Wolff says. He helped persuade Congress—which had funded NSFNet as an academic project—to allow commercial traffic on the backbone. The rest was history. With the advent of the World Wide Web and commercial browsers, telecommunication companies suddenly became eager to build new networks and hook up to the Internet. The commercial Internet was such a runaway hit that Wolff closed the NSFNet down in 1995, his last year there. He believed it was no longer needed—and being dependent on government funding, it was better for the Internet not to rely on it.

Both Wolff and Clark remain very active in Internet affairs. Wolff is the principal scientist for Internet2, a consortium founded in 1995 that manages a new academic backbone like NSFNet (only much, much faster). “Internet2 has more than 250 academic institutions as members,” Wolff says. “Every country at this point has an academic network like Internet2,” he adds, “and all are interconnected in a global architecture to support research and higher education.”

Clark is still at MIT, where he has been employed as either a student, postdoc, or research scientist for the last 49 years. He has won two Test of Time Awards from the Association for Computing Machinery. In one of his Test of Time papers, written in 2002, Clark introduced the term “tussle” to describe the interactions on today’s Internet: tussles between service providers and users; companies and hackers; governments and cryptographers. This year, Google and China engaged in a very public tussle over security certificates. The solutions do not lie in engineering alone—they involve economics, politics, and diplomacy. The best thing that technicians can do, according to Clark, is enable the tussle to play out without bringing down the Internet.

“I’ve made this transition in the 2000s from a technical person into a multidisciplinary person. I have economists and political scientists working with me,” Clark says. “Swarthmore absolutely saved me, not because of any specific thing I learned there, but because of the mindset that it’s good to talk with people who aren’t like you.”