Excerpt from a broadcast of "Science Friday" with host Ira Flatow, guests and callers
Ira Flatow: On October 29,1969 -- at 10:30 pm -- a couple of computer scientists were huddled around two computers spaced 400 miles apart. One was at UCLA and the other at Stanford Research Institute. They were able to send a message from one machine to the other. Even though the computer immediately crashed, what they achieved 40 years ago was perhaps the moment the internet was born! Here to recapture that moment was one of the guys in the room that night, Leonard Kleinrock. He's a Distinguished Professor of Computer Science at UCLA. Welcome! Can you tell us what you guys were doing in that room that night?
Leonard Kleinrock: Well, we were about to experiment with the first connection between two computers on this fledgling internet and in order to do so, we needed to send a message from Machine 1 to Machine B 400 hundred miles away, as you said, to the north of us. All we wanted to do was to log in from our computer to the second computer on the internet, and in order to do that you had to type "log" and the remote computer would type the "in" for you because it knows what you're doing. So we also had a telephone connection as well as the data connection for this new network. We typed the "l" and phoned the other end and asked, "Get the 'l'?" And the response was, "Yep!" We typed the "o". "Got the 'o'?" "Yep, got the 'o'." "Got the 'g'?" CRASH! As you said, that remote computer crashed. So the first message on the internet was, "lo," as in "lo and behold"!
Ira Flatow: And do most of the computer experts now consider that to be the moment -- the Mary-Had-A-Little-Lamb moment for the internet?
Leonard Kleinrock: Well, there's a few other times when you could think about the internet coming to life. A month before that, on September 2, the first packet switch -- we now call these things routers -- was connected to that computer I referred to in our lab at UCLA over a 15' cable and bits began to move back and forth. So I'd like to say that's the day when the infant internet took its first breath of life. But it was not yet doing what it was planned to do, namely to connect computers together. It's on that October 29th date when I like to say the infant internet uttered its first words and the first message went. Just as we say the telegraph was invented when Samuel Morse sent that wonderful, biblical message, "What hath God wrought?" That sort of sequence has occurred many times in new technology emergency.
Ira Flatow: Tell us what this new technology of packet-switching was, compared to the old technology?
Leonard Kleinrock: It began when I was a graduate student at MIT. I was surrounded by computers and I recognized one day that those computers would need to talk to each other. The only network that could support communication at that time in any sensible way was the telephone network run by AT&T. It uses a technology called "circuit switching" as opposed to what we're about to describe as packet switching. In circuit switching, the problem is that it holds the connection between the two communicating parties -- you and me, for example -- for the entire duration of that conversation. Even if I take a coffee break or you decide to step out for a moment, the line -- communication path -- is still reserved for our use and it's being wasted. In voice communications that happens about a third of the time and that's quite acceptable. But with data communications, the communications path is idle perhaps 99.99% of the time. I recognized we could not afford to use such a network as inefficiently as that. So we had to find a way -- and I did as a student at MIT -- to share those resources. When you take your coffee break then, in fact, someone else can use the line. The idea of demand access is where it's only being used when you have data to send -- and when you're not someone else can grab it. The way we did that was to present this data message that you want to send to the system, the system chops it up into little pieces called "packets," and each packet hops its way through the network until it finds its destination. Each packet may take a separate path. At the other end, those packets are put together and the message is delivered -- as if a letter were chopped up into a number of postcards and each one send through the US mail system separately. The key idea here is that if there's no packet to be sent in a particular part of the path, other packets from other users can use that particular part of the path. No portion is wasted if there's some work to be done. That's called "packet-switching."
Ira Flatow: How do the packets know where to go? How do they know how to be reassembled at the other end?
Leonard Kleinrock: Each packet has an address. It says, "I want to go from New York to Los Angeles and I'm going to go through Chicago and perhaps Denver. How does it know to go through Chicago and Denver? Because there is in the network a control procedure which is constantly checking which links are giving good performance. When a packet arrives in Chicago, Chicago will say, "Well, the best next hop is to Denver and not to Dallas." So there's intelligence in the network which is constantly probing where the good paths are. It selects them on a demand basis: which one can take this packet best next?
Ira Flatow: That's why there's some delay when you're loading your browser? Trying to reassemble all those packages?
Leonard Kleinrock: Well no. The packets are reassembled at the destination. That's not where the delay is. The delay is in the speed of the line from the end of the network to your device or, more often, in your operating system.
Ira Flatow: And what does the router do? Is that what you helped invent? The modern-day router?
Leonard Kleinrock: Yes. I invented the concept of demand access, one example of which is packet-switching and along with that this idea of distributing control -- distributive responsibility for deciding which are the good paths and lets select them on a dynamic basis. Right now, where's the best way to go? The interesting thing is that when I started to look at this problem I said, "I want to build large networks, not small ones. So we cannot give control of this large network into a single node or a single location. I had to distribute that function across the entire network. So everybody's smart, everybody's looking for good paths, and they're constantly being found as traffic arrives to be resent and re[inaudible].
Ira Flatow: ...Was there a thought in people's minds in those days of actually sending email and messages as well? Or was it just for sending data?
Leonard Kleinrock: In fact, the original purpose was simply to share resources between computers so that if you had an excellent graphics processing capability and I wanted to use it, I would log onto your system through the network and use your graphics by logging on. This was data exchange. It was also, therefore, people to computer exchange. So I could reach out. What I never anticipated -- and by the way there is a written, published press release that came out before the network was deployed in which I do articulate a vision -- that it would come to be that this was a network of people communicating with each other. The social networking. Email didn't come in until 1972. Once it did, it was clear to me that this was about people communicating, not about machines talking to each other. But it took two years before that recognition was made clear.
Ira Flatow: You know, this seems to mirror the history of communications in general. The newer inventions were made for businesses. Even the telephone was not invented supposedly for people to talk to one another. It was to send data down the line or information. The idea that people would be talking on the phone and making money on that was kind of foreign at the time. Alexander Graham Bell was laughed at for that reason.
Leonard Kleinrock: There was no economic model here and it was, in fact, for science purposes. These exchanges were between scientists who were conducting research into computer science and other areas of interest.
Ira Flatow: Is there any way to predict where this is all heading?
Leonard Kleinrock: There are two parts to that question. One is, what are the applications going to be. And the best thing we can predict is that it will be impossible to predict those new applications that are coming. We've had a sequence of surprising applications throughout the history of the internet. Email was the first. The web. Google. My Space. Facebook. Twitter. YouTube. Nobody saw those applications coming. They hit us on the side of the head. We can anticipate that applications will probably arise in the domain of mobile computing, in location-based applications, in video delivery. But I can tell you what my vision of the infrastructure increase and modernization will be. I believe we're going to be able to take cyberspace out from behind the screen of the computer and bring it into your physical environment -- in the walls of your room, in your desk, in your fingernails, your eyeglasses and your automobile! And when you walk into a room, the room will know you walked in, and it will say, "Ira, what can I do for you?" And you can have a conversation with it. It may respond with voice, a hologram with a display. This technology will be made possible by employing embedded technology in your physical environment. This technology will provide internet access for you. There'll be sensors and actuators and logic and cameras and memory and display based on today's mems and nanotechnology. So we're going to end up with basically a pervasive global nervous system across this planet.
Ira Flatow: It'll be like I'm at the hollow deck on Star Trek!
Leonard Kleinrock: Exactly. And it's all going to be invisible. You won't see it. You won't have to fuss around with clumsy interfaces which we have today.
Ira Flatow: Let's take a call from Larry in Wyoming.
Larry: I was first going to ask about your perspective on Al Gore's role in helping to get the internet going, but I figure a better question for your knowledge would be if you think the US is going to be able to transfer successfully to IP v.6 or if they're going to have the same problem that they've had transferring to the metric system? Most of the world is using the metric system now. We're still using ASC or whatever -- where most of the world is using IP v.6 and the US is still using IP v.4.
Ira Flatow: What about the Al Gore question?
Leonard Kleinrock: Sure: the Al Gore question is an interesting one. Al Gore was a very important contributor to the expansion of the internet. In the late 1980's he was perhaps the most knowledgeable person in Congress. He would visit us all the time. We'd hold meetings. He understood the technology of the internet better than anyone in Washington. In fact there's a report I wrote about a national research network and I testified for his Congressional sub-committee. He convinced the first President Bush to sign into law as his final act the High Performance Computing and Communications Act of 1991 which allowed academia, business and government to cooperate to deploy a much higher speed backbone network. So his contribution was to bring additional government funding to enhance the backbone network. He did make a contribution and it was very important. But it was much later than the birth of the internet. It was some twenty years later.
Ira Flatow: So to say that Al Gore helped invent the internet would not be far from the truth in terms of keeping it going.
Leonard Kleinrock: He certainly was a contributor to keeping it going and enhancing it. But certainly he was not there at the birth.
Ira Flatow: Anything surprising that has happened over the years, something that surprised you?
Leonard Kleinrock: Yes, in fact. In those early days the culture of the internet was one of trust, openness, shared ideas. I knew everybody on the internet in those days, and I trusted them all. Everybody behaved well. So we had a very easy, open access. We did not introduce any limitations nor did we introduce -- what we should have -- which was the ability to do strong user authentication and strong file authentication. So that I know, if you're communicating with me, it's you, Ira Flatow, and not someone else. If you send me a file, I receive the file you intended me to receive. We should have installed that in the architecture in the early days. And the first thing we should have done? Turn it off. Because we needed this open, trusted, shared environment -- the culture, the ethics of the early internet. And then when we approached the late '80's and early '90's -- and spam and viruses and pornography and eventually identity theft and fraud and the botnets and the denial of service we see today -- as that began to emerge we should have then slowly turned on that authentication process. Which is part of what your caller referred to. IP v.6 is an attempt to bring on -- to patch on -- some of this authentication capability. But it's very hard now when it's not built deep into the architecture of the internet.
Ira Flatow: What about "net neutrality"? What are your thoughts on that?
Leonard Kleinrock: Network neutrality is a very key issues these days. The question is, should there be a controlling body that determines what kind of content you receive, base its pricing on what it is rather than how much communications resource it uses. I believe that the government and the corporate world should not determine what content I can receive or how its charged. I should pay for bandwidth but not for it's a video this or that. Whether I pay for the type of content depends upon the content provider and me. Except there should be some controls in the case of some proven abuse. I believe in a very open, free network except where there is abuse occurring which is clear and significant.
Ira Flatow: Call from Scott in Davenport, Iowa.
Scott: ...I just wanted to point out that I heard an interview on NPR last night with one of the gentlemen that had actually helped send the first message across the Arpanet. He told the same story -- that they did the "l" and the "o". When the machine crashed, they went back, worked the bugs out, came back a little while later and tried it again. So the first three letters sent across the Arpanet were "LOL"!
Ira Flatow: Leonard, is that...
Leonard Kleinrock: That's correct! The "lo" as the first attempt. The second attempt was to start that over again with an "log." So that is correct. You have to understand: if you think about what Alexander Graham Bell and Morse and even Armstrong did, those guys were smart. They had the public relations ready, the media, and they had a powerful message to send. We didn't. We weren't that smart! But the message we ended up sending, the "LOL" ...
Scott: ... I just think it's quite appropriate that "LOL" were the first three letters sent across the Arpanet.
Ira Flatow: And they're probably still the most popular letters still sent across the internet! Do you think we're going to get control of all these forces? That are trying to turn it into a bunch of junk?
Leonard Kleinrock: Well, right now I think the internet is approaching teenage. And it's behaving like a teenager. It's mischievous, it's unruly, it's unpredictable. The dark side is a manifestation of that. If you think you can control and modify and improve the behavior of a teenager -- as they pass from teenage to young adulthood -- the answer is an optimistic "yes." But along the way they can go wrong, and whether or not we succeed in maintaining and basically subduing some of the bad behavior is a really difficult open question. When you get the forces of the crime syndicates using the dark side of the internet, then you have a problem in how you're going to control them. It's well beyond the internet.