3/26/18
By Peter High, published on Forbes
Vint Cerf is considered to be one of the fathers of the internet, having been the co-inventor of TCP/IP, having led influential work at DARPA, then at MCI, where he pioneered an email platform called MCI Mail. Now 74, he remains as busy as ever. Since 2005, he has been Google’s Chief Internet Evangelist. It is a role that gives him plenty of freedom. As he notes herein, “I focus on doing whatever I can to create conditions under which more Internet infrastructure can be built. This means I need to go where there is not Internet infrastructure to find ways to get it built. In weaker economies, affordability is a big issue, and the rationale for investing in Internet infrastructure is not necessarily obvious.”
As our conversation continued, however, he shared several concerns about the future of his creation. He indicates that his goal is to determine how to make the internet safer, more secure, and more private. He also co-founded the People-Centered Internet (PCI) with customer relationship management pioneer, Mei Lin Fung, installing former FCC CIO David Bray as the executive director. He notes that, through PCI, he strives to identify projects that could “materially improve people’s use of the Internet.” He sites as examples, “creating information in local languages or providing services that improve people’s ability to find jobs or improve their economic stability or safety or health.”
Peter High: As the co-inventor of TCP/IP, you are famously considered one of the “Fathers of the Internet.” Can you talk about your early work and the genesis of those ideas?
Vint Cerf: Thank you for being careful about that, as Bob Kahn and I had two hands on one pencil. He deserves huge credit, as do tens of thousands of other people since that time. You do not do anything on this scale without a huge amount of willing collaboration and commitment.
When I was a graduate student at UCLA, I worked with others on the ARPANET project, which was a predecessor to the Internet. ARPANET was a packet switching experiment within DARPA to connect a dozen universities that were doing computer science and artificial intelligence research for the Defense Department. The idea was to link all their computers together so they could share their resources, computing capabilities, and results to make progress more quickly.
The challenge was how to connect those machines together since the circuit switching technology at the time would have been too slow for the interactions we needed. Packet switching, on the other hand, is more like electronic postcards. They get lost and they come out of order. You must do a bunch of little things to make a postcard postal service work reliably, and the same is true for packet switching. We got the ARPANET running, and Steve Crocker, who remains one of my best friends, led the network working group to develop the protocols to allow different brands of computers to communicate across this packet switch ARPANET.
After I finished my Ph.D., I went to Stanford to work on computer networking. Bob Kahn, in the meantime, had left Bolt Beranek and Newman and joined DARPA. He came to my lab and said, “We have a problem.” The problem was that we were going to use computers for command and control because it would help us better manage our resources. However, this would require us to put computers in mobile vehicles, ships at sea, and airplanes, in addition to fixed installations. Obviously, these could not be wired connections, and we were going to have to use mobile radio and satellites in addition to the dedicated telephone circuits that we were using to build ARPANET.
The problem was that the packet switch nets differed in terms of sizes, speeds, and delays. We started working on this problem in the spring of 1973. By September, in collaboration with the International Network Working Group, we had a solution. We began detailing the design in January 1974 and published a paper in May of that year. By the end of ’74, we had a fully detailed specification for what was called Transmission Control Protocol (TCP). The next several years involved implementation and testing and discovery of mistakes and their repair, so we iterated several times until we ended up with a final specification in 1978 which we froze.
By that time, we had split the Internet Protocol [IP] off from the Transmission Control Protocol [TCP] to deal with real-time communications that did not require reliability but required timeliness. With radar communication, for example, you do not want to know where the missile was, you want to know where it is now. You do not need to retransmit old information because it is not of any use. We split the protocols into TCP and IP, and we created something called a User Datagram Protocol which gave the users access to this real-time communications channel.
That occupied my time by 1978, at which point I was already in the Washington area working for the Defense Advanced Research Projects Agency (DARPA). We continued implementation across as many operating systems as we could and on January 1, 1983, we turned the Internet on. At the time, it consisted of approximately three networks: the ARPANET, mobile packet radio net in the San Francisco Bay Area, and a type of satellite net over the Atlantic. This was followed by rapid growth in the academic community, which DARPA supported.
Without going into another 20 years of history, the National Science Foundation (NSF) picked up the idea and funded the creation of the National Science Foundation Backbone Network and about a dozen intermediate elements to connect 3,000 universities around the US into this growing Internet system. They contributed an enormous amount to the absolute growth of the system and made some important decisions that allowed the network to eventually become a commercial service.
High: At what point in that journey did you see the broader commercial and global implications of what you were creating?
To read the full interview, please visit Forbes