After lunch today at WTF 2004, there were several short, sharp presentations in which speakers were limited to a four-minute time slot. What follows are partial transcripts of the first round of talks, which addressed open-source telephony, investing in innovation, the telecom industry’s pricing strategy, and other topics.
Why Don’t Venture Capital Funds Invest in Innovation?
Bart Stuck is a partner in Signal Lake, a venture capital firm that invests in early-stage companies working in the broadband telecom and networking infrastructure space.
I spent eight years at MIT and 12 years at Bell Labs. Then, in a five-year period I made a 760% internal rate of return on a series of investments. The premise of the venture fund was that we would go after innovation. Most of the money we put in is in wireless. What we discovered was that there was not a lot of interest in funding innovation. When Bell Laboratories was spun off with Lucent in 1996, we felt that there was going to be some hard times for basic research if Lucent fell on hard times.
We looked at 1,300 IPOs between 1993 and 2003 in computers and communications. We looked at 200 mergers and acquisitions. And we ranked them in terms of innovative capability. We came up with five companies that were truly innovative. We came up with Verisign, Sierra, and others, and we aggregated all the data. We were puzzled that so few of the companies that went public and were truly innovative got venture funding and actually went someplace.
What actually happened was that those companies ended up with higher valuations than those that got the venture capital. And when you had the crash, those valuations held up. The other companies that went public had valuations that were not as high and were merger candidates. We’re puzzled over the huge influx of money in venture funds — $3 million in 1993 to over $15 billion in 2002. Is the issue life cycle? Is the issue systemic? Is the issue the people in the venture funds — MBAs and bankers?
Telecom Economics in the Post-Glut Era
Stan Hanks was formerly the vice president of research and technology for Enron and currently works with Columbia Ventures, a Washington state-based organization that buys up bankrupt telecommunications companies.
In the late ’90s, we saw everyone building networks all over the place, and they all came from the same place. Qwest built a large network and sold two thirds of it. Wall Street loved it. We had a bunch of people building more capacity than we could ever use. In 1998, I built a network from Salt Lake to Dallas. It cost $110,000 per mile. We sold it for $1,500 a fiber mile. That network now is selling for $100 a fiber mile, and there are no buyers.
Those are the economics that are in play. It’s actually salvage; it’s the junk business. We can get huge CAPEX savings from buying assets out of chapter 7 or 11 — or 18 or 22. It becomes not terribly negotiable, and the real estate can end up killing you. It can cost me $5,000 every time I roll a truck. That adds up over time. The other thing is that you’re not done with capital expenditures when you think you are. You need to get hardware recertified, and you’ve got some hardware obsolescence. Then you’ve got fitness of purpose. If your core business shifts you’ll need equipment that you don’t already have.
There are some smoking deals out there. If all you want to do is buy glass, you can buy a whole lot. But the purchase price isn’t the whole story. It’s still not actually possible to run some of these assets profitably. There’s a theory that some of this stuff is worth buying now because it’s so cheap. Problem is you’ve got to pay to have it sit there. My prevailing rule of thumb is that if Wall Street really loves something, run like hell. That’s what got us into this.
Music at Bell Labs
Steve Crandall used to work as a member of AT&T Bell Labs’ technical staff.
About 10 years ago, several of us at Bell Labs were doing a lot of fiber optics research. What became clear is that how people use these things is a lot harder than just the network. Since then we’ve gotten together with multidisciplinary crews to try to work out these problems. The one that’s most interesting to people is music. There are about 10,000 societies in the world that you can characterize linguistically. All but two employ music in the form of singing.
We started dealing with conservatories and places like Oberlin College. How do you combine this with the network? When my wife had her 50th birthday, what do you do for a present? My friend, who was in voice at Oberlin, said my wife likes doo wop. She knew some people who performed doo wop and composed it. On my wife’s birthday, she got a phone call, and people performed a doo-wop number that was written just for her.
How do you get that experience on the Internet? Things like the iTunes Music Store are just incremental interim solutions. If you hang out with students, you learn a lot. I have seven years of data from four colleges, continuous data monitoring of 10,000 students. If you look at the usage of MP3’s, 71% are listened to less than once. 14% are listened to once. People use this for sampling.
There are people described by ethnographers as having a library engine. On campus now, everyone has their iTunes players on. They’re using Rendezvous to share up to 120 music libraries. Your social status is based on your placement in that list of libraries. What iPods are useful for is when the networks aren’t big enough. There’s also the phenomenon that you can help people get past the stage of the hardness to make music. Soon, people will spend more money on making music than buying music. Garageband lets you make really, really bad music. But you’re making your own music.
How do you deal with this path issue? Want to make the Net really interesting to me? To make me want to have fiber in my home? The path between me and good music at Oberlin is three meters. In central New Jersey, it’s a kilometer.
Bruce Kushnick servces as chairman for Teletruth, the alliance for customers’ telecommunications rights. He’s also one heck of a keyboard player.
This is about us, what we want out of the network, and what we’re never going to get if we let the monopolies continue to control things. The Bell companies were formed in 1984 when AT&T was divested. The Bell companies fell on their knees and said we’re only little companies, we need more money. So they started offering new services like call waiting that cost pennies to operate but charged $5 for it. It was 1991-1995 when the Bell companies promised to roll out the Information Superhighway. The Bells were really clever. They said they couldn’t build the networks because they didn’t have the money. They introduced price caps. They’d freeze rates, lay off some staff, and use the leftovers to build the networks.
They did this in New Jersey, Maryland, and Pennsylvania. Customers spent thousands of dollars on networks that they never got. The telcos pocketed the money. The new trend is for the companies to say they still can’t roll out the networks without more money. The law has said that we’ll let the Bells own the networks, get rid of line sharing, and kill off the ISPs.
The entire telecom crash was caused by the telecom companies that did not roll out the networks they said they would before 2000. Had they done that, we’d have high-speed lines, new computers, new software, new services, and the last mile. How do you get people annoyed by this? They’ve killed off all the little local ISPs. Verizon is the creation of GTE, Nynex, and Atlantic. We all know according to science that siblings should not marry.
We, the rebel alliance, need to take the networks back. We pay $17 billion a year for networks we will never get. How do we get the customers to understand that? Let’s go after the phone bills. All of those little charges can’t be correct. The thing that scares me the most is that if the Bells get control of the network and the cable companies, they’ll kill off things like VoIP. We need to figure out how to keep the network common carrier. And we want gigabit speeds.
Aperiodic Grid: Stupid Software
Tom Mandel is a poet, entrepreneur, and pioneer in social computing. Ken Tyler is the proprietor of 8th Fold, the maker of Seed Wiki, a wiki farm. While Mandel presented the following, my understanding is that it’s the outcome of a collaboration.
The hardest thing to grasp is that we in this room are the ones who wish to build railroad tracks along the canal to haul the barges. We are talking about everything but anything that users want or have asked for. This is a four-part matrix that represents the conversation we’re having. We’re talking about demand and supply, and we’re talking about it from a perspective of either anxiety or comfort. There’s been more anxiety than comfort. But there’s nothing about what people do, need to do, want to do, or have done with others.
Any network, you can look at in terms of its center and edge, and whether it’s stupid or intelligent. This schema can be reproduced for other applications. Only people are intelligent. Nothing else is intelligent. Devices are dumb. You unplug them, and they just sit there. They’re dumb. With the computer, we created an interface with folders, files, and a pointer. Now we have much more computing power, but we still have that interface. So we’ve built in a whole lot more intelligence.
That’s putting intelligence in the center between me and what I’m doing. All intelligence that’s not real intelligence is competing with the intelligence of people. This morning, Eli said “I don’t believe you can run an entire economy on niches.” That’s missing one word: “except.” I don’t believe you can run an entire economy except on niches. A society is based on niches, the local. Economy is a function of creating society. A state can declare that a town is not an entity. You make that true by eliminating a niche.
This creates not only anxiety but a kind of repetition and a treatise on anxiety. You just have to look at what the value propositions are. There’s no limit. The value propositions that we’re being offered aren’t real. The offerings that we need respond to perceived social needs and perceived niches.
Open Source Telephony
Michael Olson is president and CEO of Sleepycat Software, maker of Berkeley DB.
I’m going to assert that open source software is an important engine of innovation in telephony and telecommunications in general. But the most interesting open source telecom company is doomed.
Many of our customers are building telecommunications systems that are handset and voice bound. Their services are data management intensive. The reason is twofold. First, the software is cheap and cheap is good. Cheap software builds an inexpensive platform. That’s a very important step. Stuff is getting commoditized, which allows us to attract new investment. It’s not that commoditization is a bad thing; it creates new opportunities higher up.
There’s a project called Asterisk, which is an open source PBX. That software is mostly written by Mark Spencer. He makes a living as a company called Digium. That’s a mistake. Digium exists to sell hardware so you can deploy a PBX on your own. Hardware is an extremely difficult business. He can make software available as services and make a relatively good business on his own. He’s not able to do different levels of service. What he has chosen to do is compete as a hardware vendor.
Digium is an interesting product and an interesting company, but it won’t be here in a year.