Meet the Professor Who Was There When the Internet Was Turned On
I'm at UCLA's computer science department, inside 3420 Boelter Hall, and in front of me is the internet's very first node—essentially the ancestor of today's routers.
Source: S.C. Stuart
This is where—on Oct. 29, 1969, at 10:30 p.m. PT—the first ARPANET message was sent across a remote network for the first time using packet switching protocols. Stanford's Research Institute was the recipient.
Standing next to me is Dr. Leonard Kleinrock, a Distinguished Professor of Computer Science who developed the mathematical theory of packet networks while completing his PhD at MIT in the early 60s. He was there on that day in 1969, when they turned the internet on.
Dr. Kleinrock told PCMag about the ARPANET days, how he got the geek bug and a full scholarship to MIT, as well as UCLA's new Connection Lab, which will open in 2019 thanks to a gift of $5 million to honor his achievements. Here are edited and condensed excerpts from our conversation.
The internet, as we know it, started in 1969, but you were developing the theory of packet switching at MIT much earlier than that.
Yes, I first published a paper that introduced packetization and mathematically evaluated the performance of packet switching in April 1962. I was looking at network design, and I wanted to evaluate how a data network would behave. Among other things, I started looking at priority—how we ordered the messages and how that affected performance.
We realized early on that you don't want a short message to wait behind a long message; that's a big delay for the little guy. If you reverse it and favor the little guy, the big guy doesn't care that much but how do you figure out which are the little guys? Every message will claim to be tiny. So you say I'll give you a little bit of time and if your message gets through in that time, you've proven you are small and you'll get done. If not, you'll get to the back of the line and wait for another bit of transmission time. That's what's known as the round-robin procedure.
Dr. Leonard Kleinrock, a Distinguished Professor of Computer Science at UCLA
But this was all purely theoretical at the time. You didn't have a packet switch dependent network to test it out on?
True. It was a mathematical analysis but I successfully simulated it on the TX-2 computer at Lincoln Lab so I knew it would work. The problem I posed to evaluate network performance was a really hard one for two reasons. One, the queuing theory part of it was, and still is, unsolvable, and the networking part you couldn't solve because it was a multi commodity flow problem—and you still can't solve that. So I introduced what is now known as "Kleinrock's Independence Assumption" and that assumption broke apart the queuing problem allowing me to analyze it. To test the assumption, I simulated the network on the TX-2 and sure enough, the results with and without the assumption matched up very well.
Would you say there was an atmosphere of discovery back then?
Oh, absolutely. The idea was in the air, it was going to happen. On the computer science side, we saw the challenge of early internet design and protocols as solving an engineering challenge. There was no notion of proprietary technology, intellectual property, ownership, or money. None of that! We got our "jollies" from doing a good job, having someone else say "I like what you did and I'm going to use it." That was a golden era at MIT and elsewhere in the research groups in the sixties, and I'll be forever grateful to ARPA's enlightened funding culture.
Take us back to 1969. Describe what it was like that day and what happened.
Well, there were only three of us there that evening in late October. There was no fanfare, no cameras, no voice recorders, no nothing.
Yup. Having said that, earlier on September 2nd after the IMP [interface message processor] was delivered and turned on, right after the Labor Day weekend, and we connected it to the host machine—an SDS Sigma 7—the room was packed. AT&T, Honeywell, Scientific Data Systems, BBN [Bolt, Beranek, and Newman], ARPA, UCLA—all waiting to point the finger at the other guy if it didn't work.
Record of the first message ever sent over the ARPANET.
Not much has changed in tech there then—"We couldn't replicate that issue, it must be their machine."
By summer 1970 there were 10 nodes across the US. Do you recall why those locations were chosen, strategically?
It was all about the people—the research being done in various locations as we expanded the network in those early years—from Xerox Parc to Stanford, Utah to London and out to Norway. And how we were all connected to each other via ARPA's funding structure. It was an amazing time.
Tell us about using the very early internet to send an "illegal" message.
In 1973 we convened for a meeting on computer communication networks at the University of Sussex in Brighton, England, and we stayed in dormitories on the campus. For some reason I had to leave early and I left my electric razor behind. When I got back to L.A. I thought, "I wonder what crazy person would be logged somewhere on the network 3 a.m. UK time?" I thought such a person might be Larry Roberts. So I used the Resource Sharing Executive program to look for him. I typed in the command: "Where Roberts" and it logged onto every host, looking at the "who" list, to see who was logged on until it found Larry Roberts.
When it located him, it gave you a message back?
Exactly. It said: "Roberts logged on as TT [teletype such-and-so] at BBN," so I connected to him with a "talk" session. Next day, I got my razor back.
Dr. Leonard Kleinrock, UCLA, 1970
Let's get your backstory now. Your first interest was in radio while at the Bronx High School of Science, right?
Ah, no, it was long before that. I was a typical kid on the streets of New York. I loved gadgets, puzzles, games, and comic books. Inside the Superman centerfold, one day, was a set of instructions which explained how to build a thing called a crystal radio.
Without being impolite, what year was this?
Probably 1941; I was 6 or 7. It claimed you could hear music by building something out of parts you could get in the street and it wouldn't cost a penny, you didn't even need electricity, no batteries or nothing.
How did you fashion the crystal?
I made the crystal out of pencil lead and my dad's razor blade. But then it said that you needed an earpiece to hear the music. There was a candy store down the street that had a telephone so…
You ripped off the candy store to get your ear piece?
Don't tell anyone.
Hey, it was in the pursuit of science.
[Laughs] Then I also needed something called a variable capacitor, so my mother took me down to Canal Street, in NYC, and I walked up to the first store, and bought one for a nickel. It was an amazing experience making that crystal set and hearing the music. I was hooked after that on engineering, science and technology.
Interface Message Processor (IMP) at UCLA
What happened next?
I took the exam for Bronx Science on a whim and got in. But then onto City College—night school—because I had to work in the daytime and bring money into the house, to contribute. So I got a job in an industrial electronics firm on Nassau Street during the day and went to school at night. But all my teachers at City College were also working during the day, they were in the field, that's so important. I always say that to my students today—you have to know what's going on, you must read the trades, you must get involved.
Great advice. So how did you get to MIT from there?
One day someone from MIT was coming to talk about scholarships to MIT and so I took off work two hours early to attend the lecture, after which I went to the professor at the back of the room and I said: "I'd like an application" and he said "Who are you? What's your name? I don't recognize you." I told him I attended evening session and he said "Get the hell out of here" and wouldn't give me an application.