It was far too many hours past midnight, yet the MIT graduate student was still chipping away at his Ph.D. dissertation on the computer. Yes, the computer. In 1962, MIT’s Lincoln Laboratories only had one, and it took up the entire basement. The TX-2 cost millions of dollars, so the grad student was careful not to make any wrong moves that might cause a malfunction and set the lab back thousands in repairs. But a high-pitched noise was interrupting the machine’s quiet hum. Panicking, the grad student started to inspect each module, trying to figure out which of them was going to bring his pursuit of a doctorate to a dead end. He slowly let his gaze wander up one of the panels, only to find … a set of eyes staring right back at him.
The eyes belonged to fellow Ph.D. student Larry Roberts, who was watching his classmate from an empty slot in the computer rack. Roberts broke out into a huge grin and just kept going: beeeeeeep.
Roberts was good at that kind of prank, mostly because nobody expected them from the naturally stern-looking young man who was known for cutting through all the nonsense and getting things done.
It was the latter ability that gave Lawrence Gilman Roberts, who died Dec. 26 at the age of 81, a place in history. Because one of the things Larry Roberts got done was the internet.
In his dissertation from 1963, Roberts developed the basics of computer-generated 3D graphics, some of which are still in use today. After graduating, he continued the work at MIT Lincoln Labs, but eventually he soured on graphics: “I knew I was 30 years out from anybody being able to use it seriously, so I started losing interest,” he told me in an interview in 2012.
It was around this time Roberts reconnected with former MIT professor J.C.R. Licklider, who had taken on a new job funding computer science research through what is now DARPA: “Licklider and I began talking about how knowledge spreads and civilization develops. We knew that the development of language helped knowledge spread through centuries, and print had done it in months. But we needed to do it in seconds to move civilization forward,” Roberts said. Licklider had a vision of an “intergalactic network” for fast knowledge sharing, but before he could implement it at the Advanced Research Project Agency (the D for Defense was added in 1972, creating the more-well-known DARPA), he went back to teaching.
Licklider was succeeded by Roberts’ MIT classmate Ivan Sutherland. He gave Roberts money to explore whether two different computers located as far from each other as Santa Monica and Boston could connect using phone lines and if this could be the basis of a network. The computers connected fine, but there was too much noise on the lines to make a stable data connection.
When Bob Taylor took over the computer research office at ARPA, he insisted that Roberts should come to D.C. to build the “ARPA network” or ARPANET, threatening to pull Lincoln Labs’ ARPA funding if he didn’t. In January 1967, Roberts begrudgingly agreed and called in all the best brains he knew, from classmate Leonard Kleinrock to the designer of the TX-2, Wes Clark.
By fall, the group had sketched out the first network consisting of several computers of different makes and models. But Roberts still hadn’t solved the problem of those noisy telephone lines. He could use dedicated phone lines, but that meant paying for them even when not in use—and risking massive congestion when they were. Roberts found his solution at a computer conference where a group of British scientists introduced him to a concept called packet switching, introduced a few years earlier by Paul Baran at the RAND Corporation. With packet switching, the data messages would be chopped up into smaller chunks (or packets) and sent through the network individually. They would then be reassembled at the receiving computer to once again form a coherent message, having individually utilized all the best and fastest pathways through the network to get there.
Packet switching was such a great concept in theory that nobody believed it would work. Experts from AT&T and IBM warned Roberts against wasting federal research funds on something that couldn’t possibly become a reality. But it worked.
After a year of work in conjunction with the engineering firm Bolt, Beranek and Newman, the first ARPANET connection between computers at UCLA and the Stanford Research Institute happened on Oct. 29, 1969. ARPANET eventually became a huge success, but packet switching as a concept made an even bigger difference. It worked so well that almost all the digital communication we use today is packetized. Your smartphone’s wireless data connection. The digital cable channels on your TV. And of course, the internet—it’s all packets.
Even email came out of ARPANET. Bolt, Beranek and Newman engineer Ray Tomlinson made electronic messaging on a single computer work across the ARPANET in 1971, using the @ sign for addressing. But Tomlinson’s email system wasn’t very useful until Larry Roberts created the first real email client, complete with an inbox that collected all incoming messages and folders you could save them in. Only then did email truly become ARPANET’s killer app. Once again, Larry Roberts was the man who made it happen.
Roberts, ever resistant to laurel resting, realized that even with packet switching, the ARPANET had a huge limitation: It was landlocked. Combining his ARPA checkbook and his network expertise, Roberts began working with researchers at the University of Hawaii to build the world’s first wireless packet network, ALOHAnet, which came online in June 1971. The ALOHAnet technologies have since found their way into cellular telephony, Wi-Fi, and even into Ethernet network cables that are by definition not wireless.
ALOHAnet and ARPANET were separate systems until, after a bit of prodding by the ALOHAnet researchers, Larry Roberts oversaw the interconnection of the two networks via satellite in December 1972. This was the first packet-based “inter-net,” though nobody called it that in 1972. Still, the cat was now out of the bag.
Almost a year earlier, Roberts had persuaded Bob Kahn, one of the engineers who worked on ARPANET at Bolt, Beranek and Newman, to come join him at ARPA. Kahn wanted to build on the ALOHAnet achievements by creating two separate packet networks using satellite and radio, respectively and internetworking them together. This would enable data to flow from land-based computers across oceans to cars driving on the roads of another continent. Kahn enlisted the help of another ARPANET alum, Vint Cerf, and together they created the protocols that still make the internet run today: TCP/IP.
As one of his last acts at ARPA, Roberts threw his support behind Kahn’s internetworking project. As soon as it was underway, he left ARPA and passed his leadership baton to Kahn.
Roberts joined Bolt, Beranek and Newman to start the first commercial packet data network open to the public, Telenet. It seems odd now, but until then, no one had considered charging consumers for use of a computer network. In 1979, Telenet was sold to a company that would ultimately become the telecom giant Sprint.
Why, you may ask, have you never heard of Larry Roberts if he was so influential in developing and implementing technologies still in use five decades later?
First of all, no single person is the “mother” or “father” or the internet. Hundreds of people contributed to its emergence, even in the early years. Second, we unfortunately often confuse achievements with likability and communication skills. The people lauded in the media as internet “fathers,” such as Vint Cerf, Leonard Kleinrock, and Tim Berners-Lee, have all made invaluable contributions. But they’re also excellent storytellers and charming personalities. The soft-spoken Larry Roberts and his stern demeanor didn’t work well on TV, and he spoke in technical facts rather than sound bites. But Roberts, driven by his credo of faster knowledge sharing for civilization’s sake, did more than anyone else to get the job done. Where others theorized, Roberts executed.
Like that time during the ARPANET design phase when he built a card-counting mechanism into a cast and, pretending to have a broken arm, went to Las Vegas with Leonard Kleinrock. The pair were determined to beat the house at as many casinos as possible. Instead, they got banned from several establishments.
As always, the resistance to Larry Roberts’ system arose because the system actually worked.