What everybody is trying to do is re-create the old AT&T, the vertical AT&T. What you need is the last loop, the backbone, and the ability to offer services [like TV, telephone and Internet] across the whole thing. That's the golden goose. There are certain regulations preventing that, but they're in the process of going away. In the next three or four years you'll see a tremendous amount of consolidation as people try to put things back together.

The thing that the communications companies like is that these services are sticky. Long distance isn't sticky. You can change your service supplier every day, and nobody would know. But your local number is as sticky as it gets. Nobody wants to change that. Your e-mail is fairly sticky, too. So they're trying to tie the rest of their services to that last, stickiest step, which is that last local piece of the connection. That way you're using our services, you're using our backbone, and AT&T can come in and say, "I'll sell you a wireless phone, a permanent fax number, I'll sell you a voice-to-fax translator, I'll sell you an e-mail address, I'll sell you a local telephone, I'll sell you Internet access, an entire bundle of stuff. That's what all telecom companies will try to do.

So do you see the regional telephone companies as your biggest competitors?

Absolutely. They are the other people with wires in the home and so they are the competition. All the other stuff is noise until they figure out who they're lined up with. In the end how do you reach out and touch that final consumer? All the other stuff -- Yahoo, AOL -- is meaningless until they team up and figure out how to exploit the asset of reaching out to the final consumer. Everything else is transient. The churn rate on this stuff is tremendous. You're not going to turn off your local phone and you're not going to turn off your Internet connection when it's on broadband in your home. Our churn rate is nothing.

As a consumer, I'm not sure I like the sound of that stickiness. Why should I want it?

The way the system works now is that I have a phone number at home, a phone number at work, I have a phone number in my car, a phone number on my handheld phone and a beeper, all of which have voice mail. I have to call five different numbers to check my voice mail. I don't want to do that. That means I want to have one number. But if I have one number that's pretty sticky. I don't want to change it.

Linking all the services sounds OK to me. But why can't I have that and still be able to switch carriers without changing my phone number?

I think technologically one day that will be possible. It's not today. There doesn't exist the technology to take your number and move it somewhere else. But with IP that will happen. When everything goes IP you will have an identifying number for wherever you are in the network at any given time. As a consumer I think that would be a good thing -- but that still doesn't mean you wouldn't buy a bundle of services from someone who can offer you a bundle of things. It costs me less money, because I only send you only one bill. Bills cost a buck or two bucks each, and if I'm offering five different services, I'm probably saving eight bucks, and I can offer you a better rate.

How far ahead do you look in your planning?

The vast majority of our planning is within a two-month horizon. It's pretty rare to have the luxury of actually being able to sit back and know what this will look like in a couple of years. Just about the only time we do that is once a year we go through a major yearly planning cycle where we plan the budget for the entire year and that oftentimes causes you to reflect on where you're going over the next two or three years.

It's very different [from Silicon Graphics]. We commonly worked on projects that were two or three years in scope. When we started a new initiative it was generally within a three-year horizon to when that product line would be developed. You had six months in the planning cycle, 18 months to two years in the engineering cycle and then another six months in the manufacturing and integration. And then it would hit the market, and it would live in the market for another two to three years, and then it would be replaced by the next thing. Here, that would never work.

The Free Software Project
Read Andrew Leonard's book-in-progress on Linux and open source -- and post your comments.