Interview: Ethernet's Inventor Sounds Off

So, what about Bob? Engineer-scientist, early Internet developer, Ethernet inventor, entrepreneur, pundit, a man who eats his own words — and, now, a venture capitalist: Bob Metcalfe has seen it all in 40 years on the front lines of engineering and technology. From his perch at Polaris Ventures, above Boston's high-tech corridor, Metcalfe recently sat down with EE Times executive editor Patrick Mannion to chat about his investments in optical, wireless/RF, hybrid fiber/coax and supercomputing companies. Along the way, he waxed eloquent on the future of the Internet, the rise of the blogosphere, the demise of print media, why engineers should not become venture capitalists and how to solve the energy crisis. Fasten your seat belts.

EE Times: What technologies excite you right now?

Bob Metcalfe: Anything to do with video and anything to do with latitudes and longitudes. GPS is spreading, so companies that produce, consume or manipulate information containing latitudes and longitudes are a big thing. It's exciting.

I'm looking at my first energy deal right now. I'm determined to get the world out of its current energy mess. I don't like being beholden to monopolists in foreign countries — or to the crazies currently running the environmental groups.

EET: What are the options?Metcalfe: Better [gas] mileage would help, nuclear would help. Removing pollutants would help, carbon sequestration would help.

EET: How many companies have you invested in as a venture capitalist?

Metcalfe: Five. First was Narad [100-Mbit/second service over hybrid fiber/ coax], then Ember [ZigBee], then Paratech [tunable circuits for RF front ends] and SiCortex [a supercomputing company designing a dense Linux cluster]. The fifth is Mintera [optical transport gear].

EET: How do you see the role of the VC?

Metcalfe: I began with a conceit that I have since tempered. VCs split their time between choosing companies and helping companies, and my conceit was that since I had built a successful company [3Com], I was going to be helpful. What I've learned over the past five years is that it's much more important to choose, rather than to be helpful. You have to do both, obviously, but if you're going to generate a big return for your investors, which is our job, then you have to be sure that you choose very carefully, as there's a limit to just how much help you can give a company.EET: Did you expect rapid turnaround, and has that happened?

Metcalfe: There was a period during the bubble when rapid turnaround was the rule. Now it's the exception. We plan on five to seven years between initial investment and liquidity.

EET: Why did you become a VC?

Metcalfe: I had been a high-tech publisher and pundit for 10 years, and I was finished. I wanted to do something else but wanted to stay in the innovation business. Ten years is about the right time to do something new. So, I published my book, Internet Collapses and Other InfoWorld Punditry, based on columns I wrote over 10 years chronicling the rise and fall of the Internet. Then I went into the VC business, as it allowed me to stay in the innovation business.

EET: What's the premise of your book?Metcalfe: Just a collection of columns with some fierce rebuttals. The title comes from a series of columns predicting that the Internet would collapse in 1996. I actually quantified the nature of the collapse and was only off by a factor of eight — but I ate the column anyway. [In a stage appearance, Metcalfe put the column in a blender with some water, and ate the resulting mixture with a spoon.] For many people, I'm better known for having eaten my column in 1997 than for having invented the Internet.

EET: What did the book say about the Internet's future?

Metcalfe: I had written a column on the seven or eight ways in which it would collapse; one of them was security, and that's certainly true. And then spam.

EET: Are there solutions to spam?

Metcalfe: There are two. One is economics, the other is permission. These were left out [of the original Internet], and now we're having to retrofit. Putting some economics around e-mail would greatly diminish spam. Permission means you can't send an e-mail to someone without their permission. And since e-mail should be strongly encrypted at all times, e-mail filtering is really the wrong way to go, since you shouldn't have third parties reading your e-mail. The best option is a combination of economics and permission.EET: What about security?

Metcalfe: I wrote the first RFC [request for comment] about how the Internet is vulnerable to security attacks. It was a warning. We forgot to put security in, and one of my pieces of advice is that anonymity should not be the default, as it is now. Source-field packets are not inspected routinely, and they should be. There should be ways of achieving anonymity in certain circumstances, but it shouldn't be the default. It's the cause of spam, the viruses, the worms and the denial-of-service attacks. All rely on anonymity.

EET: Is that being addressed?

Metcalfe: No. It requires that the router vendors start inspecting source fields, and I believe some of that has already started — I think Juniper's routers inspect source fields. But no one turns it on.

EET: Why not?Metcalfe: Maybe they're not as smart as I am.

EET: Do you still predict the collapse of the Internet?

Metcalfe: I never predicted the fall of the Internet. What I predicted were collapses, which are outages. I even quantified it: I predicted a "gigalapse," or 1 billion lost user hours in a single outage, and the biggest one that year was 118 megalapse — pretty close to a gigalapse, but not quite, which is why I ate the column.

The Internet remains fragile, but it has inherent resilience. My dire warnings were somewhat exaggerated. The Internet is getting better all the time. We're moving to video this decade — video mail, videoconferencing, video-on-demand and video merchandising. Our children are currently stealing CDs, but they'll be stealing DVDs soon.

EET: What do you see as the most interesting development today?Metcalfe: The blogosphere is perhaps the most interesting thing going on. I'm watching the blogosphere dismantle old forms of journalism. I'm watching the daily newspaper go down the tubes, as it so richly deserves. The New York Times laid off 200 people here in Boston recently and I celebrated that event, with the [Boston] Globe and the Times being corrupt and perverted. It's just an amazing coincidence that the publisher of The New York Times just happens to be the son of the previous publisher. His son just happens to be the best-qualified, left-leaning [candidate]. He just doesn't understand what newspapers are for. He thinks they're his personal propaganda machine. But the readers have caught on and they're reading it less, and he's laying off while the blogs are blossoming. And I see that as a beautiful future. They provide choice, and freedom and competition and multiplicity.

EET: But there's no editing.

Metcalfe: Some blogs will get edited and filtered over time. Reputations will evolve and variations will evolve. Let a thousand flowers bloom. The secret to progress is an acronym I have called FOCACA, standing for freedom of choice among competing alternatives.

We're learning how to search and filter. Google searched Web pages and now we're developing the ability to better search the blogosphere: connecting facts, opinions and information better.

EET: Initially there was hope the Internet would open up intersocietal communication and eliminate borders, but some see it as having enabled more closed groups, with like-minded users feeding among themselves. Is that a reversal?Metcalfe: That's rubbish. The Internet is spreading freedom around the world and that's a very positive thing. Look, all the people in the NYT have gone down a rat hole together. I'm not suggesting [the paper] be shut down — just that everyone cancel their subscription.

Print isn't going to die suddenly, but it is dying. Be careful. Still, it'll take a long time. The [computer] displays are getting better, etc. However, keep in mind the four B's: beaches, bathrooms, buses and . . . I forget the fourth. You can't bring your computer there. On the other hand, I guess you can, now.

EET: Ethernet has changed since you invented it in 1973. How would you define Ethernet today?

Metcalfe: Ethernet has evolved considerably over its 33 years, and the word has lost its original meaning. I've often given the speech that the one enduring quality is the business model, with six features.

It's based on a de jure industry standard; second, the implementations of a standard are owned by companies — vs. the open-source model. Third is fierce competition among vendors, it drives progress; fourth is that this competition is not based on incompatibility. Interoperability is required by the market so the buyers can choose vendors. Fifth is that the standard evolves based on market interaction — meaning rapidly. And sixth is that no matter how rapid this evolution, there's a high premium based on backward and forward compatibility. This is the most enduring part of Ethernet.EET: What is the relationship between Ethernet and your work now in ZigBee?

Metcalfe: Ethernet was proposed, and became the solution for, networking PCs. ZigBee is the proposed solution for networking embedded computers. Ethernet began as standard [IEEE] 802.3, ZigBee began as [IEEE] 802.15.4. So, they both have standards. Also, there are the protocol stacks that go on top: For Ethernet, it eventually became TCP/IP, and for 802.15.4 it's proposed to be ZigBee. There's an argument about that, as there was about TCP/IP on Ethernet, so for .15.4 it's still an open question, though ZigBee is the leading contender. Others include Millennial Net, Crossbow, Dust [Networks], Zensys and then the whole TinyOS world. ZigBee is the commercially supported control standard.

EET: Does ZigBee have the same forward- and backward-compatibility strengths as Ethernet?

Metcalfe: Yes. Right now Ember believes that the 15.4 [standard's] 2.4-GHz radio will be the most important medium, but not the last medium. There will be magnetic, subgig[ahertz] radio and UWB [ultrawideband], so the stack has to evolve to assume a variety of media below it. There will also be some accommodation for more profiles as more and more applications are developed for ZigBee. There's still a debate over whether to put IP [Internet Protocol] over an 802.15.4 radio for monitoring and control purposes. That's a respectable debate, and if the addresses weren't so damned long it might even win the day.

EET: What are ZigBee's chances vis--vis the competition, and how will it evolve?Metcalfe:If you look at the industrial control markets, one of the markets ZigBee is in, it is a very fragmented industry that traditionally was based on proprietary lock-in. It's tough selling into that market, since the notion of a standard is not automatically accepted — in fact, it was fought by the vendors. Ethernet was rejected because of its collisions, since they said they needed guaranteed real-time response. Ethernet didn't offer that, and they were allowed not to buy it. You hear a lot of those same arguments now aimed at ZigBee. "We have to be wired, mesh won't work, it's too nondeterministic."

EET: Magnetic, UWB, 2.4 GHz — which will ZigBee sit upon?

Metcalfe: I think they're all going to come. Our investigation indicates that 2.4 [GHz] will have the most potential for the next few years.

EET: Where do you sit in the 100-Gbit vs. 40-Gbit optical debate?

Metcalfe: I need to disclose that I'm an investor in Mintera, which sells 40-Gbit/s ultralong-haul optical transport equipment to carriers. So, in this debate I'm riding a particular horse. The carriers have tended to upgrade their networks by factors of four. Hence the move from 2.5 to 10 to 40 Gbits/s. However, for the Ethernet world, we generally go up by factors of 10. We're now at 10G, so the next logical step is 100G. But the laws of physics do apply, and it's harder to go to 100 than 40. So, the question is, will [100-Gbit Ethernet] be worth it, and what will the price points be, etc.? So, we're proposing and selling 40G to carriers, where you can actually send 40G over 3,000 km. We just demonstrated that. The Ethernet world often goes just 10 feet, so maybe you will get 100G there. Of course, since we're all using IP, the packets can fly at 100G in one place and 40 in another. So, that will allow the network to have a mixture of speeds.EET: Is 40-Gbit Ethernet regaining momentum?

Metcalfe: Mintera managed to survive the telecom meltdown, as it was wisely managed. It cut its burn and went into hibernation. We decided to invest in it again a few months ago, because we felt the Internet was continuing to grow and there was a measurable filling up in the optical fiber out there in the carriers. So we figured in about three years — though we're not sure when — the interest in 40G would pick up. Evidence is that the 40G [uptake] is increasing.

EET: Any advice for someone wanting to become a venture capitalist?

Metcalfe: Yes: They should not become a VC. There's already too many of us. There's a thousand firms, there's too much money. Go into medicine — or become a blogger or an artist or a doctor. But don't become a VC — that's my job.

EET: Are you still the same bombastic Bob, or have you mellowed?Metcalfe: If anything, I've gotten even more opinionated. Like a lot of other people, I lament the quality of discourse around the world. But as long as people I disagree with are going to be nasty, then I suppose I have to be nasty back.

Robert M. Metcalfe

Born:
April 7, 1946, Brooklyn, N.Y.

Education:
Massachusetts Institute of Technology, bachelor's degrees in electrical engineering and in management, 1969; Harvard University, MS in applied mathematics, 1970, and PhD in computer science, 1973

Career:
Polaris Ventures, general partner, since January 2001, specializing in Boston-area information technology startups

InfoWorld, 1990 to 2000 (CEO, 1990-95): Wrote an Internet column for eight years read weekly by more than 500,000 information technologists.3Com Corp., 1979-1990: Founder and, at various times, chairman, CEO, division general manager and vice president of engineering, sales and marketing

Xerox Palo Alto Research Center, 1972-79: research staff member, computer science, and then manager of systems architecture

Engineer-scientist, 1965-1972, at Raytheon, Adams Associates, MIT Lab for Nuclear Science and MIT Project Mac

Author:

Packet Communication (Thomson), Internet Collapses and Other InfoWorld Punditry (IDG Books) and Beyond Calculation: The Next Fifty Years of Computing (co-edited for Springer Verlag)

Most recent awards: