Internet Buzz with Richard Wiggins | 5

East Lansing, Michigan
An Interview with Vint Cerf - Part 2

By Richard Wiggins

r. Vinton Cerf is known as "the father of the Internet" because of his key role in developing the Internet Protocol (IP). Currently Cerf is a vice president in the Internet division of telecommunications giant MCI, soon to be spun off as part of the Worldcom merger.

Recently, prior to presenting a keynote address at the University of Michigan, Dr. Cerf answered questions in a press conference. With him was Dr. Douglas Van Houweling, who leads the Internet2 and Abilene projects. Here is Part 2 of our interview. (See also Part 1 of our interview with Vint Cerf.)

Now, there is another interesting motivation for the expanded address space. It has taken about 25 years to get the Internet where it is now, from the beginnings in 1973. Although the rest of the world may not see it in those terms, I sure do. And, looking 20 years ahead, I'm optimistic we may see colonies on Mars and even research stations on some of the moons of Jupiter. And I would like very much to have Internets running in those facilities, and able to connect with each other. So we're now, we're – "we" being Vint Cerf and a bunch of guys at JPL (the Jet Propulsion Lab) – designing an interplanetary Internet. For that, you need some serious address space, because we don't know how big each of the planetary Internets is likely to get. So we're I think quite serious about needing more address space than we have today – and we'll probably need it sooner than any of us really appreciates.

You have some interesting latency problems with interplanetary internets…

Absolutely. We're working on an interplanetary gateway that would behave differently than today's routers. So we have an Internet of internets, again with layered protocol, and the interplanetary Internet gateway will have to be doing some application level conversions. We're not going to run TCP all the way to Titan or something like that – an eight hour round trip or something like that.

And you'll get that technology from the Talon Center or where? [The Talons are an advanced civilization from the television series "Earth Final Conflict," for which Cerf serves as a technical advisor. – ed.]

Actually if I could get the interdimensional thing going [laughter] it'd save a lot of trouble, because TCP would work in that case! [laughter] Unfortunately we haven't figured out how to do that yet…

In a more mundane and terrestrial application, one of the things we're working on in the Internet2 project is what we call "tele-immersion," where we take two researchers and allow them to sit in what feels like the same laboratory, even though they're separated by several hundred, maybe thousands of miles from one another. Using haptic feedback – you know, gloves – they can actually work together on a molecule and try to get together and say "no, Joe, let me try this – see, it's got to be like this" kind of interaction around a space.

There, we actually have latency problems terrestrially…

Van HOUWELING:

…because once you get above 30 or 40 milliseconds of delay, this whole "haptic feedback" – which is what virtual reality people call this – when you touch something, you feel the pressure from it – stuff's not gonna work for human beings – and if people are on different coasts of the United States, you're already beginning to push on that pretty hard…

Well, your 30 milliseconds one-way delay -- the speed of light in a fiber is not 186,200 miles per second, it's only 90,000, so round trip is 60 milliseconds. And that's enough time so that you twitch it this way, and the other guy has been twitching it the other direction suddenly feels you tugging on it, and it's very awkward…

And so one of the things that folks who work in this area are now beginning to do is try to develop some computer algorithms that are somewhat predictive about the likely continuing motion of the first person, so that in fact you're anticipating a little bit what's actually going to be needed, and sort of smooth this whole thing out.

I've got a great story to tell about that. Back in the earliest days, when I was at Stanford, like 1972… Remember Pong, and Space War – they each had spaceships that would chase each other and shoot. We tried to build a network-based Space War. And the problem is, you were running two different computers which each was displaying what it thought was the current state of the world, right? And you'd shoot at the other guy, right? And you thought you'd blown him up, but the other guy would move out of the way before the missile caught him on his display.

OK, so now we have a disparity. And we never quite got that resolved, until the system started making exactly those kinds of predictions – trying to get a resolution on whether or not you blew the other guy up.

You know, delay is the enemy of interactive computing.

In fact, we have a complaint to the Europeans, that the speed of light is faster over there. [laughter] Of course, that's not true, but the distances are shorter, and it's easier to do some of these sorts of things in that…

…setting…

… than it is to do it in the United States.

That's interesting. Algorithms that work for them might not work any more. It's a different example of scaling – geographic scaling. Interesting problem.

More...