The most profound observations about the forces afoot in the digital world were made by Gordon Moore, Bob Metcalfe, and Ronald Coase. Mr. Moore forecast the phenomenal pace at which computers would increase in power and decrease in price. Mr. Metcalfe showed how advances in computing would lead to networks like the Internet that, once they reached critical mass, would become irresistibly powerful. Mr. Coase explained how the kinds of changes being created by the digital economy could alter the very nature of firms.

By the standards of our fast-moving world, their ideas first appeared a long, long time ago. The earliest was published in the 1930s. The most recent appeared in 1980. So, over the summer, Context Editor-in-Chief Paul Carroll interviewed the three men to get an update on where they see the forces they identified going from here.

The answer, for better or worse, is more of the same. A lot more of the same.

Mr. Moore—whose name appears on Moore's Law—sees computer chips increasing in power at pretty much the same mind-boggling pace for decades to come. Mr. Metcalfe—author of what has recently become known as Metcalfe's Law—says the Internet will for many years keep increasing in power and value at its recent pace. Mr. Coase, who won a Nobel Prize in economics for discovering the idea of transaction costs, was the least-inclined to speculate. But he did see enormous disruptions because of the pace of change predicted by Messrs. Moore and Metcalfe. And he delivered a sort of punch line for the series of interviews. He said that the best way to cut transaction costs is to manage information better—presumably taking advantage of computer technology, which will keep benefiting from the Moore and Metcalfe effects. Oh, and Mr. Coase said those transaction costs, those cost-saving opportunities, could amount to as much as $4.5 trillion a year in the U.S. alone.

The interviews begin with Mr. Moore, one of the seminal figures in the history of computing. Mr. Moore, now 69 years old, co-founded Intel and was its chief executive until 1987. He also made two astonishingly accurate predictions about the growth of computing power, which together became known as "Moore's Law." Stated narrowly, all he said was that the number of devices on a piece of silicon would double every year or two. But that idea had incredible power. First of all, it just kept going and going. While there were just a few dozen devices on a chip in the '60s, there are a million times that many now. Second, all those doublings in power came at virtually no increase in cost. A square inch of transistor-laden silicon costs about the same now as its predecessors did more than 30 years ago. As Mr. Moore's visitors approach his "office" at Intel headquarters in Santa Clara, Calif., to begin the interview, he spots them coming and pops out. His office, it turns out, is a cubicle that is no different from the hundreds of others on that floor. The unpretentious multibillionaire, who still comes to the office once a week, has taken space down on the second floor. He does have a view, but it's of the parking lot. With no assistant in sight, he tries a conference room door to see if it's available, and the interview starts.

CONTEXT: To begin at the beginning: Where did Moore's Law come from?

GORDON MOORE: In 1965, for their 35th anniversary edition, Electronics magazine asked me to predict the progress of semiconductor components for the next 10 years. I looked at the few integrated circuits we had out at that time and what we were about to introduce and saw that the number of components on integrated circuits had been about doubling every year. We were up to 60-some at that time. I took a very naive extrapolation and said, OK, we'll keep doubling every year for the next 10. In 1975, we'll have some 65,000 components on a chip.

I wasn't trying to be accurate. I just wanted to get across the idea that we were only at the beginning of what we could do in integrated circuits—in particular, that the cost of electronics was going to drop dramatically. The amazing thing was that my prediction turned out to be really quite precise for that 10 years.

Then, in 1975, I gave a talk where I said, Look, it's going to slow down. From here on, we may only double every two years instead of every year. We were a little bit ahead of that for a while, but that's about where we are now.

Since then, Moore's Law has been applied to a variety of things I never said—though I'm perfectly willing to take credit for them. In particular, it's been used to explain the exponential growth in the performance of computing. Now, that's a direct result of the increasing number of components on a chip, but it isn't something I ever really plotted.

I think there's been one real change in Moore's Law. In the beginning, it recorded history. Now, it's what drives the industry. All the participants realize that they have to keep up with the curve I described.

The Semiconductor Industry Association has even put out technology road maps that are based on continuing this exponential growth. We're using those to try to be sure that the research gets done to let the industry continue to advance at the rate it has in the past.

CONTEXT: I've seen articles about various developments that could supposedly either slow down the pace of progress or even speed it up. As the person who developed the law, where do you come out?

MOORE: Well, I think the developments you're referring to are required to just stay on the curve we've been on. It's hard to speed up. If anything, every step gets more difficult.

We're working with feature sizes that are so small, they're hard to imagine—you could say that the features are about the size of a herpes virus, but I don't think that helps a lot. We currently use visible light to etch components on the semiconductors, but now we're getting down to wavelengths for which essentially no materials are transparent. You can't make lenses any more.

We're looking at three major alternatives to go beyond what we do now—X-rays, electron beams, and something called extreme ultraviolet, which actually came out of the Star Wars program. All have significant problems. But I think there's enough effort being put into them that any one of them could be made to work.

The next problem we run across is the fact that materials are made out of atoms. I don't see a way around that one.

CONTEXT: When does that problem kick in?

MOORE: We'll get to atomic dimensions in four or five generations of technology beyond where we are now. Generally, a technology generation is about three years. So, that gets us at least through 2010.

But, even when we can't go any smaller, we haven't really stopped the progress. That's a point I don't think I've been very good at getting across. We'll be putting multiple billions of transistors in a microprocessor. The room that leaves for the design engineers to be creative is really quite fantastic. Also, we can make bigger chips. Chips have been growing relatively slowly in size, and there's no fundamental limit other than economics on that one.

The number of transistors on a chip might keep doubling every two years. It might double every four or five years. That's still a rate of progress that no other industry approaches.

CONTEXT: Three years ago, you wrote a paper saying you weren't sure how much longer Moore's Law would last. It sounds like you're more optimistic now. Is that right?

MOORE: Probably a bit. A while back, I thought X-rays would have to take over because of the problems with the optics we've been using for so long. But here we are in production with equipment two generations beyond that and in development of another two beyond that which we can see doing fairly straightforwardly with optics.

My biggest concern a couple of years ago was that the economics were going to wipe us out. The cost of each new generation of plant was growing exponentially, along with the number of components on the chips. But, with a lot of these problems, once you recognize them you come up with a way of handling them. We actually spent a smaller percentage of our revenue on capital expansion the last few years than we did earlier. We have a real need for leading-edge production capacity and traditionally haven't had much use for equipment that was one or two generations old. But we told our engineers that their next generation of process had to be able to use at least 70% of the old equipment. We've actually been able to use 80% or 90%. Brand new capacity is still pretty darn expensive, but it's no longer prohibitive to move from one generation to the next.

CONTEXT: Your successor at Intel, Andy Grove, once told me he doesn't try to think about technology any more than 10 years out because it all feels like science fiction. Do you try to envision what the uses will be of all this additional power you're expecting?

MOORE: If Andy can look 10 years out, he's doing awfully well.

First of all, if you asked me to predict the important uses of semiconductors in the 1980s, I would have missed the PC. If you'd asked me in 1990, I would have missed the Internet. But I will say that I think we're still pretty early in how much computing, particularly combined with broad-bandwidth communications, will change the way industry works. I'll also say that electronic commerce will grow by leaps and bounds.

As for new technological capabilities, my favorite is good speech recognition. I want a machine that understands in context, just like a person does, so it knows whether I mean "to," "too," or "two." I think that when we get to that point it's going to completely change the way we interact with computers. There's an awful lot of software between now and then. So maybe this is 20 years out, but I think it's absolutely going to come and have a really big impact.

CONTEXT: Are there particular industries that you think of as changing rather dramatically because of technology?

MOORE: I think nearly all of them have. The changing of industry is ridiculously rapid. Last year, well over 80% of Intel's revenue came from products that weren't introduced as of Jan. 1. That's the only thing really that is constant: how fast the change is. Imagine if the chemical industry got into that kind of situation. Chemical plants currently have something like a 40-year life cycle. It would be awfully hard for the industry to adjust.

CONTEXT: How do you go about looking out there to see what technological change might disrupt some major industry?

MOORE: I don't have to do that any more. (He smiles.)

Mr. Metcalfe first made a name for himself 25 years ago when he invented Ethernet, which has become the standard technology for linking office computers. He later founded 3Com, the maker of computer networking equipment. He left 3Com in 1990 to first become publisher of Infoworld, a computer trade publication, and then a full-time technology pundit. Along the way, he made a prescient observation that was largely ignored at the time but that has been seized on recently as a way of explaining the popularity of the Internet. He said that the value of a network isn't proportional to the number of members. It's proportional to the square of the number of members. In other words, going from 100 telephones to 1,000 doesn't increase the value 10-fold. It increases the value 100-fold. If Mr. Metcalfe is right—and he seems to be, within the limits that he describes—then you have to take the constant doubling that Mr. Moore describes and amplify it with another exponential as you think about how quickly information technology's effects are spreading. Perhaps appropriately, given Mr. Metcalfe's belief in the power of far-flung networks, the interview takes place with Mr. Metcalfe on his car phone near Boston while Context editors in Chicago and Silicon Valley hook up with him via teleconference.

CONTEXT: Where did Metcalfe's Law come from?

BOB METCALFE: Well, it was in about 1980. I had just formed 3Com, whose principal idea was that the limiting factor in the growth of information technology was that things couldn't be connected to each other. Even when companies considered establishing networks, they wouldn't go far enough. Companies might try e-mail, which was very controversial at the time, but they would maybe have five people in the trial. After some number of months, they'd conclude that e-mail wasn't very useful.

I theorized that e-mail wasn't very valuable unless it reached enough other people so that you used it all the time. In general, I felt there was a critical mass phenomenon. Once technological standards emerged, networks would grow to a point where they would become irresistible. But I knew that, up until that point of critical mass, you had to take it on faith.

To make my point as I was out promoting Unix, TCP/IP, and Ethernet as emerging standards, I made up this little black-and-white slide that had two lines on it. One was a straight line, going up at 45 degrees. It represented the cost of the network as people were added. The other was a quadratic curve, X-squared. It represented the value of the network as people were added. Until enough people were on the network, the cost exceeded the value. But after critical mass was reached, and the lines crossed, the value would grow much faster than the cost. I argued that the crossover point was hard to quantify but was real.

Then, in 1996 or something, George Gilder called my idea "Metcalfe's Law" and began to talk about it. By the way, I now call my idea Metcalfe's Law. I'm not going to fight George Gilder over that. If he wants to call it Metcalfe's Law, then sign me up.

CONTEXT: Had you done any particular historical research about the behavior of other networks, such as the telephone, fax machines, or cable television?

METCALFE: I used all of those—in particular, the telephone.

Now, this law is either optimistic or pessimistic. It may be optimistic as the number of people on a network gets very large. I mean, when you have hundreds of millions of phones in the world, well, most people can't call 200 million people in their lifetime, so the value of the network probably doesn't continue to grow as fast as the square of the number of users.

On the other hand, my slide overestimates costs. At the beginning of Ethernet, 25 years ago, its cost was about $5,000 per connection. Now, it's $50.

Metcalfe's Law isn't a law the way Moore's Law is. Moore's Law is amazing because it's actually true. It's accurate. My law is more of a general observation.

CONTEXT: Where do you think the Web is on that quadratic curve of yours? How much more value will it continue to get from increasing numbers of people coming on there?

METCALFE: Well, the Web is way past critical mass. Critical mass occurred shortly after the 1993 introduction of Mosaic [the browser that became the basis for Netscape]. But there's no evidence that I've seen of saturation. I think we're in that portion of the curve where Metcalfe's Law might actually be quantitatively correct.

There may be other factors to consider, too. For example, value may not depend just on the number of people. It may also be expressed in what you do with those people. So, in other words, there is a certain value if all you're doing is e-mail. There might be more value if you're doing group conferencing. There may be even more value if you have video.

CONTEXT: At one point, you were quoted rather famously as saying that the Internet was going to fall apart. Why?

METCALFE: What I said and what got reported turned out to be quite different after a while.

CONTEXT: I assumed that.

METCALFE: What I began warning about in 1995 is that the infrastructure of the Internet is fragile and overloaded. Where I was wrong was that I predicted a huge outage. I predicted a gigalapse in 1996—a billion lost user hours in a single outage. That didn't happen, which is why I wound up eating my column in front of 1,000 people at a conference.

I was only off by less than a factor of 10, and there continue to be huge outages to this day, although, to be fair, there has yet to be a gigalapse. The possibility is still there.

CONTEXT: How long do you think it will be before the Internet becomes robust? Pick your own standard of robustness.

METCALFE: If you're really optimistic, you'd say 10 years. There are bandwidth issues into the home that will take years and years and years to resolve. Solutions to the Internet's known architectural problems may take 10 years to be installed. There are policy issues about how the suppliers of the Internet's backbone should relate to each other. Those could stretch to 10 years.

There's lots of capacity going on-line, which is cause for optimism. But the question of robustness really depends on your expectations. One of my readers wrote me about how wonderful it was to read my column on the Web—he could download it in less than 10 minutes. I thought, "Wow! Isn't that an atrocity. It's only 700 words." He was doing a different comparison. He was comparing 10 minutes to how long it would take him to get the magazine and open it up to the page. But I know that the column should be downloadable in less than a second. That's what makes this question so complicated.

CONTEXT: Do you see any magic bullets?

METCALFE: I have recently become a believer in cable modems. My advice is that if there is cable modem access to the Internet in your area, buy it. If not, insist on it from the local cable operator. If the cable operator doesn't seem enthusiastic, then move.

CONTEXT: What else should we watch for?

METCALFE: Two points. First, a major development will be the arrival of micro money to feed commerce. As soon as we can sustain transactions that are below a dollar, and perhaps below a penny, there's a completely different level of commerce that will be created. My favorite example is intellectual property like my column. My column is supported by advertising. I asked my readers whether they would pay two cents for it. They said, 0.2 cents. Even if you multiply my 629,000 average weekly readers by just 0.2 cents, you still have a doable deal. What that means is that intellectual property or columns or opinions or publishing or whatever you want to call it that is not supported by advertising can suddenly be supported by readers. That's new.

CONTEXT: I'll bet you could even figure out a way to let people set a dial. If I had to pay two cents to read your column with no advertising, I might be able to say instead that I'd pay a penny and take half the advertising.

METCALFE: I hadn't thought about that refinement. But, yes, set the dial. Keep in mind that many readers of Infoworld read us because of the advertising. The editors tend to view it as a necessary evil. Readers don't, at least not as much.

CONTEXT: I diverted you. You had two points.

METCALFE: My second is that I think the biggest story in information technology now is what the Internet is doing to the telephone. There's about $200 billion a year in annual telephone revenues that are up for grabs, and the Internet is going after it. This is great for everybody except the entrenched monopolies. My personal crusade these days is breaking those monopolies. This is ever more important because the Internet and the associated telecommunications have become a major factor of production in our GNP. As long as that factor of production is in the hands of monopolists, all of our economy will be hurt. If you think that's bad in the U.S., you should see it outside the U.S. You have France and Germany and other places where, if you believe that bandwidth is now the emergent factor of production of the Information Age, those economies are going to go down a rat hole because those telecommunications costs are going to stay far too high. Those countries, especially France, are turning into theme parks. It's fun to visit, but would you ever want to do business there?

Mr. Coase's is a remarkable story. As a youth in England, he was placed in a school for "physical defectives" because he wore leg braces. Mr. Coase was taught, among other things, basket-weaving but received no academic training until he was 12, when he went to the local high school. While still in high school, he completed the first year's requirements for a bachelor of commerce at the London School of Economics. This left him with an additional year to fill at LSE because of its graduation rules. When he was awarded a Cassel Traveling Scholarship for that year, he decided to do some field research in the U.S. [1931-1932]. What he observed there led him to the idea of transaction costs, which would win him the Nobel Prize in economics.

Mr. Coase, now 87 years old, offers to come in to be interviewed in Context's Chicago office, which is around the corner from his home. He begins several answers by saying, "You know better than I," and demurs on one question by insisting he doesn't know enough about the relevant branch of economic theory. When asked about the Nobel Prize, he says that "in the strictest sense I didn't win a Nobel Prize at all." He notes that the prize was established decades after the others and wasn't financed by the will of Alfred Nobel. But, despite the appearance of meekness, he offers some powerful ideas for regulating electronic commerce and for using the Internet to cut a whopping amount of transaction costs.

CONTEXT: First, could you give us an idea of how large you think transaction costs are, as a percentage of the total economy?

RONALD COASE: A large part of costs of production is what we call transaction costs. And a large part of transaction costs is information costs. Getting information. Absorbing information. Nobel Prize winners Douglass North and J. Wallis made an estimate of the magnitude of transaction costs: 45% of total cost.

CONTEXT: So, if U.S. gross national product is on the order of $10 trillion, then there could be $4.5 trillion of transaction costs in the economy? And companies could do away with a lot of that $4.5 trillion simply by handling information better?

COASE: Now, who knows whether it's 45% or 40% or even 25%. The point is, you're not going to get a small figure.

CONTEXT: You've been looking at ways of reducing transaction costs. What are you doing exactly?

COASE: At the Center for Research on Contracts and the Structure of Enterprise at the University of Pittsburgh, of which I am chairman of the advisory board, we're collecting as many contracts as we can and making them available on the Internet, so people can use them as models to generate enforceable contracts quickly and inexpensively. I think we have 4,000 contracts now. That's not a very useful number. When we've got a half million, you'll begin to get something that's useful.

CONTEXT: How should electronic commerce be regulated to keep transaction costs as low as possible?

COASE: The market itself should do it. I think it's very interesting that you can only get an efficient market when you have all sorts of rules and regulations. But, at the Chicago Board of Trade, for instance, these regulations exist quite apart from the government. When you can trade. What you can trade. Settlements. Arrangements for settling disputes. That's all arranged by the exchange. The only difficulty is that the reach of private regulation is limited. People who are trading in certain commodities can make the regulations for their trades. But trading in all sorts of things is going on everywhere. And it's very hard for people who are scattered and who have very different interests and never come together to make rules and regulations. The government, then, is necessary to make certain types of regulations. But, generally, government regulations are done very, very badly. When you go to the places where the government is most in control, you find the solution is worse than can be imagined. In Peru, for example, almost all activities are illegal because the costs of legal transactions are so great. To do anything, you may be required to get 200 licenses. That's not possible. Even the buses in Lima, 80% of them, are illegal. If people start a housing project there, they do it by occupying the land, either public or private, and then building little shacks there. You do this until the Army stops you, but the Army only stops about 10% of this. They can't go around stopping everything. If regulations can be made by people who use them, it will be better.

CONTEXT: You've criticized economists for dealing too much in the abstract, while ignoring real-world factors like transaction costs. How are you trying to address that deficiency?

COASE: In biology, you study the body. You study how the kidneys and the heart and the liver and the arteries and the blood system interact, and you make progress that way. But we don't do that in economics. You just have chaps who are interested in the heart or at home with the liver. But that's going to change. I was the first president of the newly formed International Society for New Institutional Economics. We're bringing together people in all different fields. We'll begin to be able to construct an accurate picture of how the economic system fits together. The change in economics will not come by a frontal assault on mainstream economics. It will come because all of the sub-branches change over to a new way of thinking that deals with this whole picture and focuses on real-world behavior.

CONTEXT: If this new society's work succeeds, how will it help executives understand the changing world of commerce?

COASE: Businessmen can learn from experience that if you do a certain thing, it's profitable, and if you do something else, it isn't profitable. Still there's something to be said for knowing why. You need some sort of framework for thinking. A person operating a machine need not know the physics determining its performance. Yet, if you're going to design better machines, you've got to know more than the fact that, if you operate the machine in a certain way, the machine breaks down a lot.

CONTEXT: What do you think Moore's and Metcalfe's laws will do to change the way business operates?