Chances are you are reading this article online. Are you on the Internet? Or are you on the World Wide Web? Is there a difference? For our younger readers, you may feel like these things have always been around. For the rest of us who remember dial-up connections, you know this is all historically new to us in the last 30 years. How did we get here?
This story begins with ABILITY’s Chet Cooper reminiscing on a past meeting with Vinton Cerf, Vice President and Chief Internet Evangelist at Google, internet pioneer and big reason behind why we are all so interconnected. Cerf, nicknamed: Father of the Internet, transports us back in history, sharing the origins of the technology we have come to depend on today.
Cooper: We met—how many years ago? You said you’ve been with Google now for 15 years?
Cerf: Since 2005.
Cooper: I met you when you were at MCI.
Cerf: Right. I’ve been at MCI twice, but we probably met after 1994 because my second tour with them was to build two Internet networks for them, MCI Net and vBNS (very high speed Backbone Network System) for the National Science Foundation (NSF). So, it would have been after 1994 and before 2005. I guess we can kind of narrow that down.
Cooper: It might have been just ’94. So, here’s the question I had. So your wife had this device around her neck that was helping her hearing. At the time, that was relatively new. Would that have been ’94 or earlier?
Cerf: Yes. This was probably—Bernie Widrow at Stanford University was very interested in helping people with hearing problems. So, he designed a necklace that had seven microphones in it. It was intended to be an array microphone. And if you’re familiar with array radar, for example, the way those things work is that you electronically steer the detection of the signal. In this case, the idea was to steer the microphone not physically, but electronically, to capture sound, and then deliver it to her ears.
She was, at the time, wearing a cochlear implant, which she had gotten in 1996. So this means that we met after 1996, almost certainly, because she didn’t get the array, the necklace, until after the cochlear implant, which was her first implant. So it could have been somewhere between 1996 and maybe 1998 or so. That would be my guess.
Cooper: OK, that’s getting closer. Now let’s see if it was a Monday or Tuesday?
Cerf: (laughs) Right, at 8:00 o’clock.
Cooper: So, here’s what I remember when we were talking. I thought you said something about you helped develop email.
Cerf: That’s right. Although let’s be careful about credit here.
Cooper: Ok. We take Master Card, Visa—
Cerf: (laughs) The original idea for electronic mail came during the period when time-sharing was invented. That’s about 1961, when time-sharing was first invented at MIT.
Cooper: What does “time-sharing” mean?
Cerf: Ok. When computers were first built, it was one person at a time who could use them. Then they automated the process a little more so that you could submit work to the computer with a big deck of cards, and it would do one at a time. It’s called “batch processing”, one after the other, but it was still only one task at a time that the computer would do. In 1961, using a PDP-1 machine from the nascent (early) Digital Equipment Corporation—
Cooper: Good memory!
Cerf: Actually, it’s sort of about that time I was starting at Stanford University as an undergraduate, so I didn’t have my fingers in the time-sharing pie yet. But John McCarthy and others at MIT said, “Why can’t we allow more than one person at a time to use this computer? The computer is fast enough that it can serve one person, and then quickly serve another person. And then quickly serve another.”
So, sitting at a terminal, which you see all the time now, people could be interacting with the computer, feeling as if they had total control over it, while the computer was actually paying attention to people in a round-robin fashion. That’s time-sharing. Literally, you’re sharing the time on the computer.
ARPANET and The First Email
Cerf: The idea of leaving a message for someone in this time-shared machine was beginning to evolve. It was mostly just leaving a file for someone else to look at. It wasn’t formalized in the way we think of email today. Then around 1969, time-sharing is now fairly well developed, and the Defense Department decide to do the ARPANET project to tie together a wide range of different kinds of computers in support of computer science research.
Cooper: This is considered the basics of the Internet?
Cerf: This is the origins of the—the ARPANET is the first packet-switched, large-scale packet-switched network. It was sponsored by the Defense Advanced Research Projects Agency (ARPA). One of the organizations that did the actual packet switches—we’d call them “routers” today, but we called them “packet switches” then. Actually, we didn’t call them that. We called them interface message processors, IMPs. Bolt, Beranek & Newman in Cambridge, Massachusetts, developed the packet switch.
One of the guys who was working on the time-sharing systems, Ray Tomlinson, figured out that if we had time-sharing systems tied together on this network, that it would be possible to send a file from one machine to another—which is part of the purpose of building the ARPANET. So, we could share resources, whether it was remote access to the time-shared computer or the ability to send information back and forth.
Around 1971, Ray figured out that if he could just tell his computer to send his file to another machine and deliver it to a user on that machine, that we could send messages to each other. But then he had the problem of figuring out, “How do I tell the computer which computer to send it to and which person it should be handed to? Whose filing system it should go into?”
So, he’s trying to separate the name of the user and the name of the computer, and he’s looking on the keyboard to find a character which is not already used by the other operating systems for some special purpose, and the only character that isn’t already used for something else is the @ sign. Which makes perfectly good sense: user@post. And so he invents electronic mail and the way you refer to someone is to send the message to a user@the name of the host computer. That’s 1971.
Of course, all of us who were working on the ARPANET made immediate use of this because it had the pleasant feature that you didn’t have to be awake at the same time, unlike telephone calls. That meant you could work with someone who was several time zones away. You could send a message, and when that person woke up, they could get the message and answer it.
This idea of being able to work remotely over different time zones was very attractive, and it allowed ARPA to have projects that spread over many time zones, including six, seven, or eight time zones. For example, from California to London, there’s an eight-hour time difference. I was at Stanford Universe carrying on research with my colleagues at University College London using email as our primary means for coordination because of the eight-hour time difference.
Cooper: When we talked, you mentioned you were not sure where you were, but your wife was in Europe for some reason, and you were working—it was a way for you to communicate with her, since it was easier because she was—
Cerf: Well, OK. Let me get to that. Just to pick up, email essentially gets invented in 1971 by Ray Tomlinson. I don’t enter into the email picture as other than a user until 1982. At this point I am just finishing my six-year tour at ARPA. I have essentially completed the Internet project from my point of view, and it’s ready to launch. It gets turned on January 1st, 1983, after, from my point of view, a decade of work, from 1973 to 1983.
So I left the Defense Department in very late ’82 as the Internet’s getting launched to go to work for MCI. What did I promise MCI? I promised to build them an electronic mail service called MCI Mail. So that was a commercial email service, and it wasn’t even the first one. There were several others that already existed. TimeNet had something called OnTime. Telenet had something called Telemail. CompuServe had an email service.
There were several commercial email services. They were disconnected from each other. If you didn’t have an account on the same email service, you couldn’t talk to someone else. MCI Mail was like that, too. It was an isolated system, except that when we designed it, we made sure it was prepared to connect to other email services, unlike the rest of them.
When I had finished my work on MCI Mail I rejoined my partner Bob Kahn at his new company, CNRI, Corporation for National Research Initiatives, in 1986. In 1988, I proposed to the government that we should experiment with connecting commercial email to the Internet.
That would have broken a rule because the rule at the time was none of the government-sponsored backbones of the Internet could carry commercial traffic. It was an appropriate use policy that was set by the National Science Foundation. So I convinced them that they should allow this experiment to happen, which would require the commercial traffic to flow on the Internet backbone. They approved it for a year.
In 1989, I announced that we had connected MCI Mail to the Internet. As soon as we announced that, every other commercial email service—time-shares, OnTime, Telemail, CompuServe—all said, “Wait a minute! Those MCI guys shouldn’t have this privilege. We should, too.” So they all connected to the Internet. As soon as they did that, they suddenly discovered that everyone in each of those previously private email services could talk to everybody else through the Internet.
Because they had to adopt the addressing conventions of the Internet in order to interact with it, and in the course of adopting those conventions, they could interact with anyone on any other email service. In some ways, that interconnection single-handedly destroyed the commercial email business because Internet mail was free.
Cooper: Oh! You bastard, you!
Cerf: (laughs) Yeah, right!
Cerf: Shocking! Actually, it was interesting because those email services continued for quite a long time. In fact, my story is that I left MCI in 1986 to join Bob Kahn, and the commercial MCI Mail system continued to run even after interconnecting with the Internet. Then I rejoined MCI in 1994, after eight years with Bob Kahn, and took over running the MCI mail system again. It continued to run until 2003, when I finally shut it down, and the reason was that it hadn’t kept up with all of the other Internet-based email services.
Cooper: Back to trying to see if my memory has any connection to reality, I felt like you said something that included you writing some notes down on a napkin. Does that ring a bell?
Cerf: Yes, it does. It wasn’t a napkin, actually, it was the back of an envelope. Of course, this is a classic engineering thing. Whenever you’re trying to figure something out, you go have a drink at a bar and you sit down and you figure it out on a napkin.
How the Internet Was Born
In this case, one of the stories is that Bob Kahn and I were working on the Internet design. We started working together on it in the spring of 1973. And, as I remember it, we had gone to some conference together, in New York. I believe it was an Association for Computing Machinery conference. I remember sitting down in a hotel lobby, literally with an envelope, and on the back of the envelope sketching the basic functionality of the Internet, which was pretty simple to draw because it was gateways that connected the internets to each other. And then those computers connected to each network. The protocols that made all that work were eventually called TCP/IP. That’s what Bob Kahn and I had been working on for six months, from the spring of ’73 to the fall of ’73. We wrote a draft paper, which we presented in September of ’73 to an international network working group that I chaired in England.
And then we published a paper in IEEE Transactions on Communications in May of 1974 called “A Protocol for Network Intercommunication.” It laid out what the transmission control protocol was and how the Internet would work. The thing that you remember me sketching was the basic diagram of the top-level architecture of the Internet.
Cooper: Does that still exist?
Cerf: That envelope? No, I don’t think so.
Cooper: OK, let’s write one up today.
Cerf: (laughs) The worst thing is that the paper that Bob and I wrote was written over roughly a two-day period in Palo Alto at a hotel that was called the Hyatt Cabaña at the time. It’s transferred ownership. It’s since changed to somebody else, Crown Plaza, I think, but it’s still called the Cabaña. We have a plaque hanging in that hotel that says something like, “In the fall of 1973, Bob Kahn and Vint Cerf designed the Internet in this hotel.”
The irony of all this is that after writing this out longhand on yellow tablets, I handed this paper to my secretary at Stanford and asked her to type it, which she did. And then she said, “What should I do with the original longhand?” And I said, “Throw it away because we can read your typing better than the hand-written one.” The historians have never forgiven me for that.
Cooper: I’m sorry, I’m gonna have to hit you in the shoulder for that! Ready?
Cooper: That’s a great story. So, I have some of my memories intact.
Cerf: I’m very impressed. That’s a 20-year-old memory, at least… (End of Part 1)
Part 2 Continued in Next Issue….