Robert Lucky: Yes, since I've spent my career in the Bell System and AT&T, I really know very little about entrepreneurship. So I thought I'd talk about something else I know nothing about, and actually that gave me a wide choice of subjects.

But I'd like to talk in the spirit of the title of this session and of the conference on what the Internet is doing to the economics of telecommunications.

It is very difficult to understand the economics of telecom. I've never really understood what it really costs and how to optimally allocate this cost among users. Even though I've had trouble understanding this in the past, I think now, in the current environment, I'm actually losing ground, and that I even understand it less than I did before.

There are momentous things happening in telecom, and there seems to be an instability that could crack open in a lot of different ways. One company in the United States advertises that it has an optical fiber network with more capacity than all the other companies put together. Do you know what company that is? Qwest. I hear people saying, "Who?" It's a relatively new company, and suddenly they claim to have more capacity than everyone else put together.

Another new company's coming along, and they say, "AT&T is advertising 10 cents a minute for voice. We're going to give it to you for 5 cents a minute." I suppose they can do it, until they're wiped out by the 3-cents-a-minute company that comes along six months from now. As I said, there are strange things going on, and the environment seems unstable. I want to talk about that just a little bit.

First you have to realize that we're dealing with a very artificial environment historically. This is an historical situation where price is not necessarily related to cost. In the past prices have been set by monopolies regulated by political considerations, by universal service tariff subsidies, and so as new entrants come in, in an unregulated environment, almost anything can happen, because we've been dealing in Alice in Wonderland for all these years. We're just beginning to try to understand reality. A lot of companies really don't know themselves what various services in telecom cost.

Now the second thing that's happening is that, of course, we are seeing the revolution now from circuit switching -- the traditional network -- to the packet switching of the Internet. Now everyone says, let's rebuild the network and make it a packet-switched world that will be a lot cheaper for everybody. We'll put the voice on a packet switch and everyone's going to be happy and it'll be really, really cheap. And that may be true. One of the factors that's happening between packet switching and circuit switching is the location and control of intelligence. Somebody described voice on the Internet -- Internet telephony -- as a cheap network with a $2,000 telephone. You really think about this. The total investment of the network today is supposed to be something like $300 billion, but if you have $2,000 telephones and 150 million people have them, that also is $300 billion, but $300 billion in a different place. It's on the desks and the homes of America, not inside the network, and the expense is borne by the consumers. Now, obviously, this is a revolution in who pays what for what, and it totally changes the situation. So what's going on?

Now the third thing I want to talk about a little longer is the dynamics of this. I always assumed that when I did a financial analysis, the parameters in the analysis were static. I'm thinking now that the most important thing about doing any business case in telecom is the dynamics of it, that nothing sits still for any amount of time at all, and if you're thinking it does, you're going to be out of business really quickly.

The telephone network was built of equipment that was depreciated over many years. The loop was depreciated over something like 22 years. I've heard it said that the average depreciation age in the network today is about 12 years. So you're dealing with equipment and plant that they thought would last a long time, and the problem is, it isn't lasting a long time. And the new companies are coming in with new technology that's all fundamentally better and cheaper and wiping themselves out one after another. The rate of this really has to do with Moore's Law. Moore's Law really fascinates me, and you all know that it basically says that silicon doubles its cost-effectiveness every 18 months. I've been for many years trying to understand this, whether this is a fundamental law of the universe or whether it's a self-fulfilling prophecy, like everybody knows that's the expected rate of change, so that's what they have to do. That's very possible. But, it's particularly appropriate here, because it says that everything is going to change.

About four or five years ago there was a beautiful little book published, called "Einstein's Dreams." And in this book, every chapter was, well, what if time ran at a different rate? What if time ran backwards? Or what if it had different properties? What would the world be like if time were different? This encouraged me to write a little article a couple years ago on Moore's dreams. What would happen if Moore's Law ran twice as fast? So that computers were basically obsolete before you got them out of the store, and there was no stability at all? What would happen if it ran twice as slow, so that the pace of innovation slowed down a lot? What would happen if it stopped, so you could buy one computer for your life and that would be it? Or what would happen if it ran backwards, where antique computers were worth more than the modern ones, because they were really better?

I was just sort of fantasizing when I wrote the article, but now what's happening is that there are lots of things in telecom that are running faster than Moore's Law, and they're very disruptive. Now Moore's Law of course is 18 months, but in optics it's 12 months right now. I think 18 months, Moore's Law, is sort of beautiful because it seems to me that it's balanced on a lot of innovation and some stability. But if it goes a lot faster, you get a lot of innovation but really no stability at all. In the optical fields, 12 months is said to be the rate of progress. In routers and packet switching, it is said to be about 10 months, so it's getting almost twice as fast as Moore's Law. In wireless it's nine months. These are really disruptive paces, and so what's happening is that every six months a new company can come in and say, "I'm sorry, but you existing guys are stuck with the old stuff. And now we can do it twice as cheap and goodbye. We're gonna wipe you out." So one theory is that the 5-cent-a-minute voice company will be wiped out by the 4 cent, which will be wiped out by 3 cent, and then is there any bottom? And people say, well, it's going to go to a flat national rate. Maybe a flat world rate. Maybe voice will be free on data, because you won't even see it in there. There may be absolutely no bottom to this kind of thing. The disruptive dynamics of Moore's Law applied to telecom are having a profound effect.

There's another theory of Moore's Law that I sometimes have, too, and it goes to fashion. Let me give you an example. I talked to a CEO of one of these new companies a week ago, and he's of course going to build a packet-switched network and he's going to wipe out all the old circuit guys. He puts his arm around me and says, "Bob, just between us technologists, we know that there's no intrinsic difference between packet switching and circuit switching. You could make circuit switching just as cost effective." It isn't that the new technologies are all that much better, but the point is the world is working on packet technology, and the world is no longer working on circuit switching. So whatever the world's working on, that's the thing that shows exponential increase in cost effectiveness, and the other technology gets stuck in the dinosaur regime. So it really isn't that the Internet technology is so much better intrinsically. I do believe that if you gave me thousands of people to work on circuit switches, I could make fast, cheap circuit switches, but the point is that everybody's mobilized behind packet technology. Therefore this one technology of packet switching is going to be fitted to everything, and ultimately has the potential of making communications cheap and flexible.

Now just a few comments about the other end of this: pricing. I'm really one of the few people, I think, in the world who really worries about what's fair in pricing for Internet. I mean, whenever I bring it up nobody seems to really care, and they say, "It's not going to be fair, but that's all right." But curious things are going on. One of the problems for the telephone companies is that the holding times for Internet access are very long, whereas the holding times for the circuit switches in telephony were designed for three minutes a call. The design assumed that people would get off and then the equipment could be shared with other people. But people on the Internet hang on for an hour or so, and what really brings it home to my attention is that something between 7 and 10 percent of the people who are calling up their Internet service provider never hang up. Never hang up. I mean, they just stay on. "I'm on, tough. I'm staying here." And with flat-rate pricing, of course, they can do that. I'd be tempted to do that myself. I hang on for a long time and I don't want to pay any more money, so I like flat-rate pricing. And yet there's something in the back of my head that says, "Now wait a minute; somebody has to pay for the infrastructure." I ask myself what would be a fair way to charge for this? I postulate: Should I pay by the minute? And I say, no way, because it doesn't cost them by the minute. The problem is that you're dealing with an industry where the marginal cost of additional bits is essentially zero. So in economic theory you might say that the price in a competitive market is driven to the marginal cost, but if the marginal cost is zero, how do you fairly charge for bits?

And, unfortunately, the parallel that you always come back to is the airline industry. Seats go, and if they go empty, then the company loses money. The same things happen in the telecom business. If the bits go without anybody wiggling them, the service providers are losing money. So, algorithms are going to be developed, and they'll have nothing to do with fairness any more than airline seats have to do with fairness. The airlines try to extract the maximum amount of money for the capacity that they have available. In telecom those available capacities are increasing at a steep rate, because the new carriers are building tremendous capacities. The incumbent carriers, the AT&T and Sprint and MCI and so forth - maybe not quite so much MCI - did very careful planning in the past of how much capacity was needed. They built just enough so that they didn't overspend on unneeded capacity. Unfortunately, of course, they were wrong in their projections. A lot more was needed than they ever believed, and traffic today is going up at a yearly basis something like 100 to 300% growth. It's a huge growth, and the companies are betting on the continuation of that huge growth. It'll be interesting to see, though, if that growth doesn't continue, whether these companies will be stuck with over-provisioned capacity. Then the whole market would turn around and there would be a lot of unused capacity and there could be a bloody price war in telecommunications. So, it'll be very interesting. People have even proposed in the future that your packets would go into the network and bargain for the cheapest fairs, that there would be auctions for capacity, right on the fly. Each packet would carry an ability to bargain for how much it might have to spend to get through the networks.

Lots of interesting stuff. So I need to conclude and say once again that the economics of telecom have changed drastically. You can expect, I think, confidently, to see very, very cheap telecom in the future in developed nations, and I sympathize with Carol Lee's worries, because the artificiality of the prices in some of the developing nations is very, very high, for political reasons, and it'll be very, very hard to undo this. It's interesting, for example, that the new satellite carriers - Iridium for example - have to charge different prices in different countries throughout the world. Obviously, from a technological standpoint it costs the same to make a call from any place in the world, because it's the same satellite system from wherever you are. But they will have to charge rates depending on where you're coming down to the earth, and in some countries these prices are really exorbitant. Unfortunately it penalizes mostly the people in those developing nations who have the least ability to pay. The leaders in those countries should realize that they have everything to gain by lowering those rates, but they seem to be unable to do that. Anyway, it'll be interesting. Thank you.

William Janeway: Thank you Bob. Bob's focus on the fundamental disequilibrium economics when marginal cost is below average cost is something I intend to come back to at the end of this panel. That is fundamental to understanding (the time) we're living in.