There are few people in our industry that I admire more than Vint Cerf, even though we have only met on a few occasions and briefly at best. Cerf is often cited as being one of the “fathers of the Internet” but he has done probably more than anyone still living to shape the global communications systems that we all take for granted. His career as senior vice president of technology strategy at MCI, the original competitor to AT&T – is now at an end as he takes on the new job next month as Google’s chief Internet evangelist.
For a 62-year old guy, Cerf will stay busy and still wear several hats. For the past several years, he has been a visiting scientist with NASA’s JPL to try to extend the Internet to an interplanetary system. And he also has the job of being the chairman of ICANN, the organization that is presently the central governing body of the Internet. In this part of the interview, we spoke to Cerf about his role at Google. We’ll also hear about some of his personal uses of computing, and where he sees the Internet evolving.
Q: Is a job as “Internet Evangelist” at Google the world’s best job? What exactly will you do there?
A: Haven’t performed it that role yet. To the extent that I have been proselytizing for Internet commerce for many years, the answer is yes, pretty good. I am going to be what I think of as a cross between a technical evangelist and a bumblebee. I will spend a lot of time visiting labs that Google has opened around the world in places like New York City, India, and Zurich, and make sure that the ideas are spread as widely as possible around the company. And part of my job will be keeping an eye out for new applications that we can support as well.
Q: What makes Google so unique that you chose this company?
A: Have to wait and see. I haven’t done a startup, still something that I want to do. Google has helluva good technology, and has lots of smart people and young enough to know they can’t do something and still try to do it anyway. They have also successfully monetized a substantial element of Internet applications and that is remarkable and they have done really well in terms of turning network-based services into revenue-producing opportunities. On top of everything else, Eric Schmidt is a dear old friend of mine that I have known for more than twenty years, and I have admired his work and was very pleased when he became the CEO.
Q: Is Google the last company you want to work for?
A: Do you expect me to die with my boots on? [Laughs.] I expect to be around a while, but I wouldn’t rule out doing something afterwards.
Q: When was the first time you heard about Google and what was your thought about Google then?
A: My guess is about three or four years ago, don’t really recall. I had been an Altavista fan. When Google came along, it did a better job, and I made it my home page some time ago. I find it indispensable to resolve any historical or factual issues, and even use its search functions even around the dinner table.
Q: What is your goal at Google – and what is the career goal you still want to accomplish?
A: I want to contribute to expanding the functionality that they are capable of delivering to people. They are well on their way towards creating a new infrastructure. We had infrastructures with TCP/IP and HTTP. What Google is doing is creating something that people can rely on to implement things through. I am excited about the amount of computing power that they can mount and that they can do it in a distributed way. I want to help them create an even more capable infrastructure. As far as my career goal, I have had plenty of success and fun in my career, except I really want to get the interplanetary networking system off the ground, quite literally and I have permission to spend time with JPL on that project.
Q: How is Google different from Microsoft? How are they similar?
A: Google is less than a tenth of Microsoft’s size in terms of staff, so Google is a lot more agile at this stage of the game. Google has a very distributed flavor to it — people interact with each other, and schedule things in a very federated way. I find that very refreshing. Microsoft is interested in search engines just like Google is. But I think both companies are capable of delivering different infrastructures and platforms. Google is more network-oriented than Microsoft, and is looking for ways to make networked things more useful.
Q: Is search going to become part of the OS and integrated into applications?
A: Look at what happens when you use Google desktop applications. You are seeing sorting through unstructured information. Second, sometimes having rigid directory-like structures like everything in folders and having a hierarchical structure for email can be very limiting. With Google’s Desktop, you can search across folders and do richer searches that don’t have to be uniform. You can look for emails, Web pages, documents, whatever. You don’t have to search for a single class of information.
But what you really want is to have both desktop and Net-based searches, and have a seamless ability to find information conveniently on your laptop and then if not, look outside. This would make your machine more connected to the rest of the Internet world.
Q: Gates has said that “Our search API is way better than Google’s search API.” You agree?
A: I believe that APIs in the eye of the programmer. And as such, I am not a big API fan. My focus has been on networking and protocols and making things work across the network. I am much more interested in building and designing protocols that allow things to interact with each other, even if they are designed by different parties. APIs work fine as long as I implement both sides of the interface. Protocols work the opposite way so that the bits on the wire are the same and I don’t care who implemented what protocols. Non-API network-centric way with using standardized protocols has much broader potential.
APIs are often under control of a single party, protocols – if they are real standards — aren’t and so as long as you have the right interfaces they can give quite a bit of flexibility with the two parties that have decided they want to interact with each other. They can program things however they want to so that the bits on the wire work out correctly.
Q: Microsoft is all about enabling their partners and developers. There seems to be a lot more work in this area for Google to succeed.
A: Yes, embrace and extend. That is a fair statement about Google. Google has enabled all sorts of people in all sorts of ways. How do I integrate the Google functionality into my own products? I think Google is getting better and better at that. A good example is the Google maps mechanism and how it can be integrated into other apps. I am excited about that and making it available over the Net.
Q: Will Google ever become a “find” engine rather than a “search” engine – in the sense that people will be able to find immediately what they are looking for, no matter what content it is?
A: In a funny way it might already be that, such as the Google desktop. Google is smart enough to index an enormous variety of content, and as time goes on the software will become more capable of rendering content and searching through content in a variety of different ways, and that will be exciting. Think how people behave now. They have directories of files, and email, but they tend to be separately treated as different objects. It is very sensible to treat all the objects that you deal with as more or less uniform. That is part of the hope for Google. Find ways to make all of these various objects treatable in a common way and use the same tool to search through all of them. So they are already partly there for some applications to be this “find” engine.
Q: What is your favorite Google’s product and why?
A: I was quite taken with Google Earth and Google moon. There is this thread of humor running through the company – such as the Swiss cheese background when you zoom in on Google Moon – and I find that people at Google don’t take themselves too seriously, and I like that. They are serious about their work, of course.
Cerf has played a key role in the development of the Internet, and its underlying protocols, almost from the beginning. He is among the most well-known as one of the players that began the original collection of nodes among the various academic research organizations that was the precursor to today’s network in the 1970s. And while not to diminish this achievement, his more important role was in 1989 when he convinced the federal government council to open up the network to allow exchanging emails with his mail system called MCIMail.
Of course, email has changed quite a bit since those early days. “It used to be that a message with 3000 characters was considered to be a big email message. Now with video and PowerPoint attachments, 100 MB is not unusual.” Another difference between email then and now is that now we have emails with complex formatting, HTML code and “already carrying programs around. That will make for some interesting times, and we are already seeing mobile pieces of software being transported around by email.” At least, we hope we can restrict mobile pieces of software to those that we want to have transported, unlike the current situation with viruses and other junk that goes around.
Does he ever regret connecting MCI Mail to the Internet back then? “Absolutely not, it was what broke the log jam.” Prior to then, no commercial uses or users were allowed on the network. “We had to get special permission from the federal government to connect our system. And looking back on it now, this helped to accelerate the commercialization process of the Internet. We were very deliberate and did it to get commercial opportunities into place. Back then, we didn’t think the government could afford to pay for the Internet for everyone. The only other alternative was to get a self-sustaining economic engine going, which is basically what happened.”
Certainly, the Internet is a different place today as a result. “Today’s problems are different, but also more complicated. In the early days, we were fumbling around just to get the damn thing to work. I have to envy all these eight-year-olds that come over and tell me about their new Web sites and I think, ”crap, I had to wait until I was 28 to use the Net and then we had to invent it first!” The challenges and the opportunities that it creates is just orders of magnitude than from the early days and I consider it to be quite fun.”
As an example of this progress, he mentions IP telephony and Skype in particular. “Yes, Skype and SIP are part of my vocabulary. But I have mixed feelings about Skype. If you watch the calls on a network analyzer, it looks like an attack on the target network. I would much rather have SIP standards that tell us what the ports are rather than blasting around looking like you are attacking someone’s host. I will certainly tip my hat to the Skype folks because they managed to figure out how to get it all to work with very little user intervention.”
Surprisingly, Cerf’s home collection of computing isn’t going to win any awards. He has never overclocked any of his gear, and hasn’t attempted “anything funny, other than pumping everything with a gig full of RAM whenever I can”. It is ironic for a guy that started out 40 years ago with the early computers (pre-PC era of course, we are taking IBM mainframes and calculating machines) of that day. Most of his home machines are Macs of various vintages, and he does carry an IBM Thinkpad because he has to be Windows-compatible. His home wireless networks are protected, thank you very much, and about the most interesting thing he can do is VPN into his home network so he can print stuff out remotely as reminders of what he needs to do when he returns home from a trip. That is a pretty cool idea and perhaps we should have an article posted here in the future showing how to do that. Cerf mentions he has plans to instrument his house and wine cellar eventually, to be able to track conditions.
Most of the people he deals with in his professional life are email users, although there are a handful of holdouts still. He uses Microsoft Office, and becoming less and less enamored with PowerPoint in particular. He says, “Power corrupts, and PowerPoint corrupts absolutely, that is because of the constraints it places on your ability to present material. I find it a very limiting tool, you start to think in those LIMITED terms.” He mentions Peter Norvig’s Gettysburg address on Powerpoint (who turns out to be a Google employee coincidentally).
But the most impressive thing at Chez Cerf is his wine cellar, with several thousand bottles of wine, some quite expensive and rare. He keeps track of his collection via a spreadsheet application: “It is sort of daunting when you first walk in and want to find something to drink,” he said. He rejected using a true database, like so many of you that still use spreadsheets to organize and sort data. It is nice to know that one of the giants in our industry has such simple – and appropriate — tastes in software, if not beverages.
Cerf has met many world leaders, scientists, and movie stars and feels very fortunate to have traveled in these circles. He is very impressed with Bill Clinton. “I may be biased politically. I was watching him on Larry King, and the man thinks about things and is articulate about them and he listens. Of all the world leaders I have met, Clinton has struck me as the most intellectually capable. He has just an enormous range of knowledge and engages readily in conversation and debate.”
Cerf mentions a great Clinton anecdote. “We were at the White House, during the millennium evenings before 2000. I was giving a joint presentation with Eric Lander and we had about 250 people in the room. During the presentation I mentioned my wife’s cochlear implant at age 53. I spent a few minutes explaining it because it really is nothing short of a miracle that she can hear again after having lost her hearing at age 3. Now remember this is post-Lewinsky. After it is all over, we go to East Room for refreshments, Bill is surrounded by everybody as usual. But he wants to see the speech processor that drives the implant. Now realize that this is a device which is inside your head, and has a speech processor that is a little computer that my wife wears clipped to her bra and a wire that goes up to her implant near her ear. So Bill is talking to Sigrid, and she wants to show him, and so she reaches down her front and starting to pull this thing out. Instantly a bunch of aides come running over concerned that something is about to happen. Bill says I guess I am not allowed to see underwear around here anymore.”
Cerf has an unusual perspective, having been around the Internet for so long and having done so much. Several years ago he suggested public flogging for spammers. “It was a joke, did you think anyone was going to take it seriously?” Still, he is as frustrated as the rest of us when it comes to dealing with spam, although he says for him that filters seem to be working.
Despite this minor annoyance, he is still fascinated by networking technology. He has seen the Internet be transformed from the early days when it was just about protocols and research through the past decade where it was all about applications and ecommerce. And he still talks about the next series of challenges and opportunities for the Internet. “I mean, right now there are about a billion users, so we have five billion more to go.” Doing the math Cerf means the rest of the world that isn’t yet connected nor has a computer to do so. But his frustrations with spammers pale with advancing the Internet address space.
His one regret looking back on all these years of Internet innovations was “I probably would have picked a larger address space. Who would have thought that 4.3 billion addresses were too small back then in 1977?” As a result, he has been flogging the IPv6 horse for quite some time and says, “We need to switch to IPv6 so we have enough address space. I don’t want to continue to play Network Address Translation games. We are still pushing hard and I think MCI will have implemented v6 by the end of the year in all of its operations. In the meantime, we are getting the pants beat off us by China and Japan who are adopting v6 a lot quicker than in the US.” He thinks by the end of 2008 we can get past 50% IPv6 implementations in the US, but that may be optimistic given the state of progress.
Also on his hit list is to internationalize the domain name system to allow different character sets besides Roman letters. “This is proving harder than we hoped and quite a chore to get these different character sets like Chinese and Japanese implemented.”
Finally, he is all about standards and protocols, just as he began his Internet career. “We need to move several layers upward to establishing standards. Right now we have pretty good standards at the HTTP layer but not too much above that: XML is helping, Web services will prove helpful. But we need more authentication and more distributed services. I want infrastructure and digital objects and processes to have persistence, and to be able to migrate running programs and replicate or be preserved from one machine to another.”
Indeed, authentication could be the key to more advanced applications. “We haven’t found more than 1% of all the apps that we can do in a distributed networking environment. The better we are able to do authentication, the more interesting our computational lives will be. I am anticipating a lot of collaborative computation, such as for online entertainment. A lot of these games are played in a distributed way, and kids are able to hear each other and see each other’s avatars. Soon we will get to the point where the games will look quite realistic, like Forrest Gump, putting someone into the middle of a historical picture. We will see that kind of digital flexibility and eventually will make for news reports that can be faked. You could basically fake anybody. We will have to cope with authenticity in the future.”
Cerf certainly is one of the original thinkers for our times. It was a pleasure talking to him, and he still has plenty of interesting and original ideas to go around. We wish him well at Google.