Frontier Exclusive Visionary Interview for hardware, software, system related business and and academia
Frontier Journal (FJ): Frontier Journal calling from Fremont, California.
Vint Cerf (VT): Yes, well, good for you. I'm glad it isn't a call from China where it would be a terrible hour of the day, I should say.
FJ: Yeah. So, my first question would be; in the early '70s you co-authored TCP-IP protocol while serving as a faculty member at Stanford.
VT: That's right.
FJ: I would like to know if at that time you wrote a research paper on this project and also, did you envision any business prospective for the TCP-IP protocol?
VT: Well, first of all, yes. Bob Kahn and I both co-authored a paper, which was published in May of 1974. It was called "A Protocol For Packet Network Intercommunications." It was published in the IEEE Transaction on Communications in the May issue. Subsequently, more detailed specifications were published. The most complete first publication was the TCP protocol and that came in December of 1974. It was published, as I recall, as RFC-675 and it had three authors. It had myself, a man named Gilven Delleau, and another man named Carl Sunshine.
FJ: OK.
VT: So, those were the earliest manifestations of the design. Of course, since that time, IP was split off from TCP and several other agents were invented to create the Internet protocols. With regards to our expectations when we did this work, or principle objective was to provide a non-proprietary computer networking standard to that all computer vendors, who were willing, anyway, could implement this protocol and that meant that machines running different operating systems from different vendors could inter-work over a network.
The second objective was to make this work for multiple networks using packet switching, but not necessarily using the same underlying transition technology. We intended for this to work with radio and with twisted-copper, with satellites and, of course, eventually optical fiber. Intent to make the system transmission technology independent has worked very well as the Internet has evolved over time to absorb and use virtually every transmission and switching technology has been invented since the 1970s.
Our motivation, since this was a Defense Department Initiative, our primary motivation for the Defense Department was to make sure that they would have a choice as to which computer systems they would be able to network. At the time, if you wanted to have networks of IBM machines, you had to use something called SNA. If you had a network of Digital Equipment Corporation (DEC) machines, you had to use what was called DECnet. And this was generally true for most of the vendor-supplied hardware with vendor-specific and proprietary operating systems.
We didn't want the military to be trapped into having to use the networking protocols of particular vendors so inventing this network independence and a set of Internet protocols was very much aligned with that objective. Eventually, of course, this became an international standard and was supported by public Internet services. But, at the time, it was very much a Defense Department Initiative.
FJ: Yeah, because I was curious. I heard there was very clear paper on any achievement that Turner Awards made. So, my second question is that it's amazing from the first computer design in the U.S. to its full commercialization in the early '70s, and it also took another 25 years for the Internet to be fully commercialized, so I would like to know which kind of hard technology do you envision as of today becoming fully commercialized in the next 25 years.
VT: [laughter] It asks whether I have any kind of clairvoyant abilities.
FJ: In your area. In the Internet industry.
VT: Right. Well, I don't know that my vision is any better than anyone else's. I can say, though, the trends are very strong. First of all, we're seeing mobility becoming extremely important in the Internet environment and so that is putting pressure on today's design and I think forcing us to look at different ways of supporting mobile devices from the Internet clouds with some frequency. I already can anticipate some specific protocols, changes, or at least proposals that would make mobility work.
The second prediction is that, although it's been a very, very long time in coming, IP Version 6 will almost certainly be in place 25 years, individual and unique Internet addresses for the much larger network.
The third thing, which I think is reasonably predictable, is that we will have implemented and deployed an interplanetary backbone that supports space exploration and gives access to the spacecraft, both on the surface of planets and moons, and also in orbit. It gives access to those resources from the convenience of the public Internet here on earth.
In order to achieve that objective we've had to develop at the jet propulsion laboratory a substantially new set of interplanetary protocols that work or can work adequately in the extremely high-delay environment where connectivity is frequently broken because of celestial motion. So a new set of protocols that are called 'delay tone networking protocols' have been invented and, I believe, will be widely implemented at least for space-based communication in 25 years time or less.
I think a fourth prediction that one could make is that an increasing number of devices will be on the Internet, Internet enabled. Many of them will be entertainment devices, some of them may be household devices - kitchen appliances and things like that - automobiles, and some certainly to be found in an office setting. Even something as simple as a light switch may very well become Internet enabled so that many of these devices, complex or very simple, will be part of the online environment that one can manage through the map. This opens up all kinds of opportunities, I think, for operation of system management tools. You might have a lot for you who knows the devices and so on.
I think, also, we're going to see RFID chips showing up in connection with the Internet environment where detectors, RFID detectors, will be used typically very locally to capture information which can then be looked up, so to speak, in the world wide web.
So those are just a few of the things that I think are reasonable to anticipate over the next 25 years.
FJ: So, as you mentioned, in terms of the second prediction is on top of IPV6, and about these pervasive devices, so I have three follow up questions for you. Besides a lot more address space IPTV4, I'm just gathering, is evolutionary standard instead of a revolutionary one. So, what do you think about how long it will take or it will be a revolutionary IP standard going to emerge. And, if yes, when will it be and what will it look like?
VT: OK. Good questions. First of all, it's true that IPV6 is essentially an evolutionary change. It increases address space, primarily, and it also offers a little bit more flexibility for mobility and somewhat better security tools. It turns out that we can anticipate - oh, I should also say that for the interplanetary work the delay tolerance protocols that are needed for interplanetary exchange will work OK over Internet locally, terrestrially on the surface or on the spacecraft where the latencies are low.
We actually have a new set of protocols for long-haul which we call the 'Long-haul Transport Protocol'. The protocol is designed to deal with extremely long trip times and with potential disconnection because of planetary motion. So that's what the interplanetary protocols would run on top of for, literally, the very long haul.
It's very possible that we will encounter some unexpected communication requirements that will show up as a consequence of nanotechnology. I think that as we contemplate devices that are microscopic in size and which might be deployed as center networks, they might even be deployed inside of your body, or inside of building materials and inside of other appliances as communications ideas.
Already, we've seen some sensor network development which does not use TCP/IP internally, but it uses something perhaps simpler and then it uses delay and deception tolerance protocols in order to overcome connectivity problems, power problems where it costs you a lot in power to radiate anything so that you're careful about when you do it.
So I anticipate that there will be some revolutionary changes there. Also, as radio becomes more critical as it is for mobility, I think you're going to see some very interesting protocol changes that take into account power consumption as part of the protocol. When you're operating on battery power what you don't want is a wasteful protocol that requires you to radiate a lot when you're either standing still or in motion because the more you radiate the more quickly you dissipate the battery power. So, you're going to want power-aware routing protocols, power-aware retransmission protocols in order to extend the lifetime of some of these mobile devices.
FJ: OK. So, you mentioned that you are working on interplanetary Internet that's a brand new protocol and it looks like Universal Internet. Now, my question is about the social impact of Internet and the next generation Internet. Of course, human beings are always aggressive towards space exploration and that kind of thing, but in the U.S. and other parts of the world, Digital Divide is a problem. What do you envision about the social impact of this Internet?
VT: Good question again. First of all, the Digital Divide has become quite popular to talk about in context of the World Summit on the Information Society and the Millennium Development Goal. Is a problem we will solve in large measure because all of the forces that drive costs down are at work? Certainly, as we learn how to build more and it's less and less expensive to get on the Net...
FJ: Like a $100 PC
VT: Right. Also, I believe you will see competition driving communication costs down, as well until you get some flat-rate kinds of services. So that helps right away. Second, I think there are initiatives, which are underway, some of them connected with the Millennium Development Goal to try to accelerate the acquisition of computers with networking capability in other parts of the world.
And third, the extremely rapid growth of mobiles that are also Internet enabled is erasing much of that digital divide. So I think that we will see many, many more users - Internet users - by the end of the decade, which will bring us up to 3-1/2 billion, or half the world's population, online. I do want to emphasize that I'm fully aware that the mobile telephone access to the Internet is less rich in terms of bandwidth than what you typically get from an optical fiber or cable modem.
And so people using mobiles to get associated with small screen size and potentially lower data rates. But many of the applications are attractive, including things like email, will be fairly implementable, even those mobile devices. So I see the digital divide tending to go away, partly because of economic development and partly because of the learning curve on the cost of producing the equipment.
FJ: So, the realization of pervasive computing, sometimes called pervasive execution because it combines the computing communication and also consumer electronics all together, heavily depends on the success of the Internet. And the Internet is all about Connectivity and access. Right now I'm performing an interview with you through the phone and we are interfacing through voice contact. Suppose we were performing a face-to-face interview through some eye contact, which is without any data or something. So my question is do you envision some kind of eye contact or some other way of interface or Connectivity in a future generation of the Internet?
VT: Well, I hope so. Although, I can tell you that my experience with video conferencing, which now extends over a period of three decades, is mixed. The first problem, of course, is inadequate bandwidth to deliver a good quality video. We're all accustomed to seeing really clear television and, typically, you don't get that over the Internet. That is changing even as we speak. I've used some of the Peer-to-Peer programs for telecommunications in the recent past, and the quality of the video is stunningly good. So that's one point.
The second point is that eye contact only happens if the camera angles are right. And so, depending on where you mount cameras, you may or may not get the feeling of eye contact from your correspondent.
Third, if you have more than two people communicating with each other, achieving something that feels like eye contact is actually quite hard.
The fourth thing that I worry about has a lot to do with immersive methods. If you've ever had any interaction with someone using video and if the image of that person is small, postage stamp size, or some small two inch by two inch image, it's not the same as feeling like you're in the same room. So it's only when you get to the point where the images that are interacting with each other - life-size, that you get a really immersive feeling.
FJ: But how about device-to-device instead of human-to-human? I mean not just eye contact, just a similar way. Communication device to device, not through data. Anyway, my next question would be what's your interpretation on the search for both current generation Internet and next generation Internet?
VT: Well, of course, today's search engines tend to very strongly depend on actual text as a tool for doing searching and matching. What I hope will happen, although I don't have any guarantee that it will, is Tim Berners-Lee's idea of the Semantic Web will actually happen. What goes on there is the information that's online in the net will have tags associated with it that help you understand what part of speech, molecular term is in a text.
That can be very valuable if you want to particularize your search; so if you wanted to search for authors of a given name - whoever that may be, Ernest Hemmingway, for example - then when you type in 'please search for author Ernest Hemmingway', it will not turn up hits that refer to Ernest or Hemmingway without specifically saying author. That's a very trivial example, but I think in the long term, our ability to produce this tag for indexed information will realize what Tim Berners-Lee called the Semantic Web. And I see that as a very powerful potential.
FJ: So as an extremely successful technologist in business world, Could you offer us some of your career advices?
VT: Thank you for the compliment implied.
FJ: That's a fact, actually, not a compliment.
VT: The first thing I would say is that it's hard to succeed unless you really care about what you're doing, if you really love what you're doing, if you're very committed to it. If you're not committed to it, if you're just doing a job, then success is less likely, I think, than something you're passionate about. So, to the extent that you have the option to do something you really care a lot about, you should choose that.
Second, I think, I've learned from experience that sometimes there are things that you could do that turn out to be really hard and very risky because you don't want to fail. So, my experience tells me that it's the risky things that are often the most interesting and, therefore, it may be actually good advice to say that if you have a choice between doing something risky and doing something not very risky, you should take the risky choice.
I don't suggest doing anything that's life-threatening by any means, but choosing things that you might fail at, but which would get great satisfaction if they succeeded, I think is really quite useful and workable advice.
FJ: OK. Great insight. So that concludes Frontier Journal's interview with Vint Cerf, Senior VP at Google. Thank you so much.
VT: Thank you. Let me just make one small correction; I'm only a Vice President and Chief Internet Evangelist at Google.
FJ: At Google, yes.
VT: And I am Chairman of ICANN.
FJ: ICANN, yes. Thank you so much. Bye-bye.
VT: Sure, Bye.
Back to Home Page