Tribute to Larry Roberts - Under Appreciated Internet and Packet Switching Pioneer
Introduction:
Only a small handful of people can truly be called the grandparents of the Internet, but only one of them was also a pioneer in X.25 public packet switching data networks. That individual was Larry Roberts, PhD, who died of a heart attack on December 26, 2018 at his home in Redwood City, California. He was 81 years old. Larry was also a prolific technical contributor to the ATM Forum, where he developed Available Bit Rate (ABR) traffic management which, together with the other traffic classes, provided any type of quality of service which was guaranteed.
Unlike vocal evangelists like Vint Cerf, Roberts didn’t have the same flair for self-promotion as some other internet pioneers. But without Larry Roberts and his determined vision for what networks could do, the history of the Internet and public data networks would have certainly looked very different.
Larry was a modest, humble, mild manored individual who let his technical accomplishments speak for themself. He was also a serial entrepeneur in later life. After leaving ARPA in 1973, Dr. Roberts founded or co-founded a half-dozen companies focused on computer networking. He received many awards, including the National Academy of Engineering’s Charles Stark Draper Prize “for the development of the Internet”, the AFIP Harry Goode Memorial Award, and the IEEE 2000 Internet Award. Larry was a member of the National Academy of Engineering and the American Academy of Arts and Sciences and a 2017 Fellow at the Computer History Museum (see photo below).
Brief History:
Roberts received his bachelor’s and master's degrees in electrical engineering (BSEE and MSEE) from M.I.T. in 1959 and 1961, respectively. When Larry received his Ph.D. in 1963, also in electrical engineering from M.I.T., he was already working at M.I.T.’s Lincoln Laboratory, a government research facility. He soon encountered the TX-0, an early computer, and used it to build an optical character recognition program. He also did early work in computer graphics and virtual reality. He and a colleague, Ivan Sutherland, built an ultrasonic pointing device that could manipulate objects on a screen.
In the late 1960s, Roberts was manager of the Pentagon’s Advanced Research Projects Agency, ARPA (now known as DARPA (with the D=Defense). He oversaw the development of the ARPANET, the precursor to the modern internet. The ARPANET’s first host-to-host connection on October 29, 1969 from UCLA to the Stanford Research Institute were the internet’s first baby steps.
During Larry's early days at ARPA, he showed Dr. Leonard Kleinrock (UCLA) and other colleagues his sketches of the network’s topology and logic. Hundreds of drawings later, he chose a distributed network, rather than a centralized or star configuration, to serve as the backbone of the Arpanet.
Just a week before the ARPANET made its first host-to-host connection in 1969, Roberts and fellow internet pioneer J.C.R. Licklider gave a presentation to the National Security Agency (NSA) in Fort Meade, Maryland. We don’t know exactly what they told the NSA, but we do know their topic of conversation: What networks would look like ten years into the future, in 1979. Larry’s job was about looking to the future on behalf of the Pentagon, a connection to the true military purpose of the early internet that many others have disingenuously shied away from.
Interestingly, many people at universities around the country were initially very skeptical about the promise of networked computers and Roberts essentially had to talk them into trying out the new packet switching technology. Email was one of the first applications he promoted. Slowly but surely it caught on. The ARPANET was later converted into the current Internet, hence making him one of the true founders of the Internet.
Life After ARPA:
The New York Times obituary on Roberts explains what he did after he left ARPA to go work in the private sector:
Dr. Roberts left ARPA in 1973 to become the founding chief executive of Telenet, a networking company that used packet switching. Seven years later, Dr. Roberts and his co-founders sold Telenet to GTE for $60 million.
The start-ups that followed focused on flow control algorithms for internet traffic and did not achieve the same level of success. He ultimately failed to amass significant wealth from the internet.
“It’s like you created the Spider-Man character and it turned into a giant franchise, but you only got paid by the hour for drawing it,” said [Larry’s son] Pasha Roberts.
Yet Dr. Roberts remained philosophical about this aspect of his life. “It’s sad that it didn’t benefit me,” he said in the 2018 interview. “But it’s interesting, and I love the knowledge I gained from building it and using it and understanding it from the beginning.”
Larry was the founder and CEO of five telecommunications companies, where he developed many leading edge products to advance Internet capability, QoS, and reliability. He was granted 11 patents and was invited to give presentations at many conferences worldwide.
April 16. 2015 Conversation with Larry at the Ethernet Technology Summit in Santa Clara, CA:
One such invitation was from this author, who asked Larry to engage in a "fireside chat" conversation with my multi-decade IEEE colleague Geoff Thomspon. Thanks to Paul Wesling and Ken Pyle, you can watch that video conversation here.
In that revealing interview, Larry expands on the importance of X.25 and X.121 addressing which was then compared to IPv4 vs IPv6 addressing. He also discussed the limitations of TCP. After Telenet was acquired by GTE and Larry joined that company he observed that GTE (along with AT&T) had negative cash flow. Funding of operations came from the proceeds of bonds sold, which permitted them to buy network equipment. "How were they ever supposed to get out of that?" Larry wondered.
Later in the program, Larry expands on the two high growth time periods for the Internet and the evolution and demise of ATM. Much, much more...........very informative and entertaining. A truly feel good event!
Larry Roberts, PhD and 2017 Computer History Museum Fellow
.....................................................................................................................................................................................................................................................................................
Postcript:
This April 2015 Data Networking history session went very well with a smooth flow between Geoff and Larry. The Q & A was also quite good, except for a question about artificial intelligence in the Internet possibly make it think on its own. Note that software control over the network does NOT imply artificial intelligence or any thinking machines. It’s the big data/ analytics software in cloud resident computer servers that might be able to predict outcomes based on past behavior.
Glad we could give credit to Larry as the primary creator of the X.25 protocol and his leading role in commercializing Packet Switched Public Data Networks at Telenet (later sold to GTE). Telenet was way ahead of AT&T, Sprint and other N.A. X.25 PSPDN carriers. In the mid to late 1970s, his version of the X.25 protocol was accepted by what is now called BT (British Post Office then) and Orange (Transpac then). It was later enhanced by ANSI X3S37 and CCITT SG XVII WP2 (which I participated in from 1978-1985).
—>X.25 was the only commercial and international public data network from 1976 to 1993 when the Internet went public and took over (see Larry’s comments below)! Even ISDN used X.25 on both the B and D channels (Basic Rate access) for packet switching.
……………………………………………………………………………........................................................................................
Larry’s comments:
Alan, What can I say? Your summary is very valuable as virtually no-one has realized that the developed world had a more reliable standardized packet service for almost 2 decades before the Internet. As to the fireside chat, it went great as Geoff fed new questions to me whenever I stopped. It worked very well.
Another thought: The time slot of 1 hour for this fireside chat was a huge difference from a prepared speech and was a serious benefit in enabling us to cover a large amount of material over a long period of time (several decades).
Thanks,
Larry
...............................................................................................................................................................................................
From Chrisos Kolias of Orange who introduced Larry at the history session:
It is true that Larry was vastly undepreciated – even his death did not reach many media. I found out about it from a Greek news source of all places! (Christos is Greek)
Another great contribution of his was slotted ALOHA (ALOHA being one of the wireless packet data networks and pre-cursor to later mobile communication systems such as GSM.
Larry had a beautiful mind.
Best, Christos
.....................................................................................................................................................................................................
References:
https://www.youtube.com/watch?reload=9&v=pldYQXyyn5c&feature=youtu.be
https://en.wikipedia.org/wiki/Lawrence_Roberts_(scientist)
https://computerhistory.org/profile/larry-roberts/