Welcome to the New Economy
Dr. Cerf, widely regarded as one of the fathers of the Internet, said that he planned to offer a sense of where the Internet is now and where it’s going, and to raise several policy issues, including the nature of the economics of the Internet. As a point of departure, Dr. Cerf said that the following features distinguish the Internet:
It is the largest network of networks in the world.
It uses TCP/IP protocols and packet switching.
It runs on any communications substrate.
Dr. Cerf began by suggesting a parallel between the generation of electric power and its distribution over copper wires and the generation and distribution of information and computation via the Internet. It is well known that electric motors run 24 hours a day doing chores, many of which are invisible, such as powering the compressors of household refrigerators. We each are likely to depend on tens to hundreds of these motors, often without being aware of them, as part of our modern infrastructure. The Internet and computers linked to it will be of similar importance so that each of us would depend on hundreds or thousands of pieces of software that run in the background, doing invisible jobs for us. Unlike electric motors, which tap into the power of huge generators on the electrical grid, computers tap into information drawn from anywhere on the network. The consequences of this information revolution would be as great or greater than the effects of the industrial revolution.
A Brief History of the Internet
Dr. Cerf reminded the audience that the Internet is in fact hundreds of thousands of connected networks. The reason it can function is that all the networks use the same set of protocols. An important point is that these networks are run by different administrations, which must collaborate both technically and economically on a global scale.
This system was also designed to run on any communications substrate. Dr. Cert said that when he and Robert Kahn first started collaborating, they wanted to make sure that the Internet and Internet packets would run on top of any future communications technologies that might develop. Thus, they chose the simplest format they could imagine—underlying technologies that only had to deliver digital signals from point A to point B with some probability of arrival greater than zero. The goal of designing the Internet protocol to run on everything, he said, was so important that they had a T-shirt made that read “IP on everything.” This objective has remained essentially intact for 25 years. Now, however, as a consequence of creating this communications substructure of Internet protocol, people are beginning to use it for other applications that sit on top of the underlying infrastructure.
The Future of the Internet is Uncertain
Dr. Cerf said that the Internet was still in the middle of its “gold rush” period, which has several implications. The first is that this gold rush will probably resemble others in that the people who make money during a gold rush are not the people looking for gold but the people selling picks and shovels to the miners. This is what the telecommunications companies have been doing—selling the electronic equivalent of picks and shovels to people who are looking for gold on the Internet. The second is that no one knows exactly where or how much gold will be found. There are many business models and many new ideas for businesses on the network as we move through a period of great experimentation.
Growth is Rapid but Early
Dr. Cerf noted that the period of rapid Internet growth began in 1988, when the number of computers on the network began to double each year. Between 1997 and 2000, the number of dot-com domains alone grew from 1.3 million registrations to 15 million. In the same period, the number of hosts almost quintupled. The number of countries accessible on the Internet rose to 218. A few weeks before the workshop the last of Africa’s 54 countries came online. By approximate count, the number of users has grown by more than a factor of six in the three-year period.
In comparison, there are nearly a billion telephone terminations around the world, representing a much larger network. Of those billion terminations, about 300 million are cell phones and 700 million are wire-line. The rate of growth of cell phones in some countries exceeds 50 percent per year, whereas wire-lines tend to increase at about 5 to 10 percent per year. Internet use, however, has grown from 80 to 100 percent per year since 1988, outstripping even the spectacular growth of cell phones.
Patterns of Internet Use
Internet Users are Distributed Unevenly
To illustrate the broad geographic distribution of Internet use, he presented statistics from an Irish company called NUA (Figure 1).1 As recently as three years ago, most Internet users were in North America; today fewer than half are in North America. Users in Europe have doubled in the last year, as have users along the Asian-Pacific rim. However, the population along the Asian-Pacific rim exceeds 2 billion if one includes China, India, Indonesia, Malaysia, and Japan. Of these 2 billion, only 75 million are Internet users.
Even within those broad areas, Internet penetration is geographically uneven. In Japan for example, Internet access through a mobile telephone network has grown quickly, adding approximately 16 million users in the last year and a half, from a population of 120 million. In Latin America, the number of users has almost tripled in the last year, from about 5.8 million to over 13 million. In Africa the number has doubled from about 1.3 million to about 2.7 million. Most users in Africa, however, live in South Africa, Egypt, Tunisia, and Morocco. Users in the Middle East have doubled in the last year. Most of those users are in Israel, but a significant fraction is now appearing in the surrounding Arab countries.
Internet Growth is Extremely Fast
Dr. Cerf projected that by the end of this decade the number of Internet users would approach 3 billion—about half the world’s population—if growth rates continue as they have in recent years (Figure 2). He cited his own projection made a year ago of 900 million devices on the Internet by 2006, and compared that with an estimate by the cell phone industry of more than 1.5 billion devices by 2004. When he combined his own estimates with those of the cellular phone industry and others they pointed to almost 2.5 billion devices on the network in just six years. “That should scare some people,” because the current IP version 4 address space includes only 4.2 billion end points and that number is never
achieved because the address space is allocated hierarchically like the telephone system. He estimated an efficiency of no more than 60 percent in the address space allocation, which would provide room for only 2.4 billion devices by 2006.
New Challenges and Future Trends
The Challenge of Carrying New Modalities
Another factor to take into account is that the Internet is now being asked by its users to carry new modalities, such as video, telephony, and radio. Dr. Cerf recalled experimenting with voice over the Internet as early as 1975, when the backbone of the Internet carried only 50 kilobits of data per second—about the speed of a dial-up modem today. Those early experiments did manage to carry voice signals by compressing them, but the sound quality was poor. Carrying sound on the Web is common now—especially one-way sound—and carrying radio is possible, requiring only 15 to 20 kilobits per second in the lower-quality ranges; however, because the packets do not flow smoothly and uniformly, they must be buffered. As a result, sound reception is affected by small delays. Given a sufficient number of buffered packets, sound quality is good, however.
High-quality video reception of 30 frames per second on a full screen requires a capacity of about 400 kilobits per second, which is not available on a dial-up line using a modem. The limit of a typical display on a dial-up line is about a 2-inch display, delivering six or seven frames per second. For high-quality video on the Internet, subscribers will require dedicated access via digital subscriber loops, cable modem, or alternatives to cable-like multi-channel, multipoint distribution systems (MMDS). Dr. Cerf said that he now operates a network for a commercial company that delivers five megabits per second per channel to business customers who have 45-megabit-per-second access to the Internet.
Added Challenges of Two-Way Service
Dr. Cerf noted that telephony, which is a two-way interactive service, brings several challenges that are not present in one-way video or radio transmission. In telephone the delay between two parties must be as low as possible. Conversations carried over a double satellite hop are degraded by delays of at least one second of round-trip time. Internet telephony could suffer similar delays if the packets are queued, stored, and forwarded in the current fashion. A one-second delay is enough to degrade a conversation, because one party must wait for a response from the other. Squeezing delay out of the Internet requires a change in some of the packet priorities so that one packet can be placed at the front of the queue of packets. In addition, both the backbone of the Internet and the access links at the edge of the Internet must be faster in terms of transmission data rate. Progress on the backbone does seem imminent. The underlying capacity of the
optical fiber Internet is typically at least 10 gigabits per second, with a terabit2 per second predicted for the first quarter of 2001. Most of the delay in Internet telephony today is attributable to the relatively low edge data rate, especially on a dial-up link.
More Internet-Enabled Devices
Dr. Cerf said he is convinced that there would be a very large number of devices on the Internet because people are able to build hardware that is compatible with Internet protocols. In 1999, two students at the University of Massachusetts built a two-chip Web server that did so at very little expense. According to Moore’s Law, the cost, size, weight, space, and power of hardware will drop steadily so that eventually “it becomes possible to Internet enable just about any technology.” He showed a picture of an Internet-enabled refrigerator with a LCD display and access to the Internet built nearly two years ago by Electrolux in Sweden. The refrigerator augments the traditional household mags-and-paper communication system with an ability to send and receive e-mail and go out on the Internet.
He also showed a picture of an Internet-enabled picture frame that, when connected to a predetermined Web site, downloads 10 pictures, and displays them one at a time without the need to learn Windows or any other software or to log manually onto the Internet. He suggested that more and more devices that have narrow functions and that are Internet-enabled would be developed.
He then showed a slide of a product built by Jaguar International of Japan— an Internet-enabled sewing machine. It allows the sewer to download fonts and logos into the sewing machine and sew those fonts and logos into the fabric, all at consumer prices.
Other kinds of Internet-enabled devices include Web TV, Palm Pilots, and Nokia 9000s, which are three-way cell phones—a cell phone, a pager, and an e-mail station. Dr. Cerf displayed on his belt an Internet-enabled two-way pager that is able to send and receive e-mail. It has a small liquid crystal display like the Nokia 9000 and a tiny keyboard. He said that the device is so easy to use that his staff sometimes sends e-mail back and forth across a conference table during meetings.
He then returned to the example of the refrigerator and its potential for additional devices, including a barcode scanner that would indicate the contents of the refrigerator. The scanner might not read how much milk is actually in the container, but it would remember that the milk entered the refrigerator three weeks ago and would send an e-mail reminder that fresh milk is needed. If the refrigera-
tor were online, it might search the Internet for recipes for its contents and signal that it contained sufficient ingredients for, say, chicken cacciatore. Alternatively, if you were shopping, you might receive an e-mail from the refrigerator reminding you to pick up the marinara sauce. A final example, seen in Japan, was an Internet-enabled bathroom scale. This scale sends your weight to your family doctor each time you step on the scale. This information could then become a part of your medical record.
A Trend Toward Wireless Devices
Dr. Cerf reiterated that many devices would be used on the Internet, a lot of them wireless. Cell phones are developing rapidly. While traditional GSM phones in Europe only deliver about 9.6 kilobits per second—adequate for e-mail but not enough for surfing the Internet—the GPRS, or general packet radio service, will transmit data at 100 or 128 kilobits per second. Third-generation cell phones are expected to have a burst rate of two megabits per second, a substantial improvement over 9.6 kilobits. He reported seeing in Geneva last October some Internetenabled cameras with radio links, able to take a digital picture and transmit it to a Web site. Parents could take a picture of the kids going off to school and the picture could appear on Grandma and Grandpa’s Internet-enabled picture frame five minutes later.
Many radio-based systems are beginning to carry Internet traffic. Digital broadcast satellites can do multicasting; the local multipoint distribution service (LMDS), and the MMDS alternatives to cable can carry hundreds of megabits per second. Ricochet is a 128-kilobit-per-second mobile link to the Internet, a service of Metricom. Wireless local area networks, or LANs, can transmit information at 11 megabits per second for more than 1,000 feet. Dr. Cerf reported using them both at work and at home. A wireless LAN card in his computer receives an Internet address from the local server and allows him to work anywhere in the house or office. Bluetooth is a new radio technology that enables low-power radio linkage of multiple devices just a few feet apart. A desktop, a laptop, a printer, and other input-output devices can all be linked without wires.
While online devices are attractive, they all require an Internet address to send or receive an Internet packet at any moment. This increases the demand for Internet address space leading, in Dr. Cerf’s opinion, to the need for a new version of Internet protocol from version 4.0 to version 6.0. Version 5.0 was an experiment that was discarded about 15 years ago. Version 6.0 has 128-bit address space, which means a total of 1038 addresses.
Economic and Political Dimensions of the Internet
From an economic point of view, Dr. Cerf said, an estimated $6.8 trillion worth of activity takes place on the Internet or by electronic means (Table 1). The
TABLE 1 Global e-Commerce in 2004
$3.5 Trillion (U.S. $3.2T)
Rest of the World
$6.79 Trillion (20% of the global economy)
estimated global economy will amount to approximately $32 to $35 trillion worldwide by 2004, which means that roughly 20 percent of the world’s economy may be Internet-focused in four years. For the sake of economic stability, therefore, it is essential to have a highly robust and reliable Internet.
Other Internet issues are part of any discussion of the New Economy. These include policy-related issues that will be brought into sharp relief as the Internet expands globally. For example, the Internet is not sensitive to international borders. It flows about the world without awareness of national boundaries or any of the laws that apply within those boundaries. One border-related issue that has aroused debate over the last several years is the issue of whether cryptography can be exported. The United States was quite restrictive in its policies at first. That approach has changed an important step toward ensuring the high-quality cryptography needed to ensure the security, authenticity, and privacy of electronic transactions on the network.
Who Owns Digitized Information?
Trademark, copyright, intellectual property, and patents are all-important issues on the Internet because it is so easy to move information around in digital form. The legal actions against such Internet firms as Napster arise from the concern of music companies that copyrighted material is being improperly copied or transmitted through the network. Internet service providers are not equipped to monitor the transmissions that go through their networks because of the high volume and the partial, packet-sized view of the material that flies past at speeds of a terabit a second. Even if a provider could see the contents of a packet that began, for example, “Call me Ishmael,” and could identify those words as the beginning of Moby Dick, the provider would still not know if it is copyrighted material or whether the sender was authorized to send it to the receiver. Such problems, he suggested, must be addressed at a higher level.
Who Should Regulate the Internet?
Other issues have to do with the regulation of telecommunications. Until now, regulatory frameworks for television, radio, and telephony have been distinct. They will soon have to collaborate or merge in some awkward fashion because all forms of communication can be carried over a single medium. This suggests that regulation should no longer be based on the service being provided but on whether the medium that carries the service is a monopoly. For example, cable would be considered a monopoly if only one party is allowed to use it. In the case of the twisted-pair telephone system, if only one party were allowed to use the line it would have a monopoly on the medium. As a result, society might be tempted to regulate the medium without necessarily regulating the service that is on top of it. It might not be necessary to regulate telephony, video, or radio services because the regulatory challenge lies at the core of the transmission system.
Who Should Tax the Internet?
An especially contentious issue is taxation and whether local, state, and national governments should be permitted to tax business transactions on the Internet. Dr. Cerf suggested that Internet transactions should be taxed to the extent that they mirror commercial transactions of the “real world.” Otherwise there is disparity between companies that sell books in the physical world and those that sell the same thing in the Internet world. If 20 percent of the world’s economy does depend on the Internet any time soon, and that portion of the economy is not taxed, the money to run governments will have to come from other forms of taxation, presumably from income and property taxes. This debate has yet to be resolved. 3
The Internet in Outer Space
As a final issue—and one that will have an impact on the economics of the Internet—he predicted that over the next 20 years the Internet would expand into outer space. “And this is not a pipe dream,” he said. “This is real engineering activity.” He described working at the Jet Propulsion Laboratory for two years with a team that is designing an interplanetary Internet protocol. The design is essentially complete, he said, and the first prototypes have been implemented.
For a discussion of the debate over taxing Internet commerce, see Hal R. Varian, “Taxation of Electronic Commerce,” Internet Policy Institute, April 2000 at <http://www.internetpolicy.org/briefing/4_00_story.html>. Professor Varian points out that, because online commerce is global, its tax treatment will also be global and must deal with countries that employ value-added taxes as well as those that use state sales taxes. Within the U.S., he writes, “The current system of state taxes is overly complex and poorly designed. No matter what one thinks will happen with online purchases it stands in need of serious reform.”
The team will insert an Internet protocol on a satellite to be launched at the end of 2000 around the Earth; another is scheduled for a lunar landing a year later and a third protocol system is scheduled to be used in 2003 on two rovers that will explore Mars.4
The larger objective is to “create an Internet backbone in space.” Each successive mission will carry a portion of that backbone and, to the extent that those assets have lifetimes beyond the mission itself, each mission will increase the size of that backbone toward the eventual goal of a major interplanetary backbone that will support the exploration of the solar system in the near term and, in the longer term, experiments in the commercialization of space. If launch costs can be pared down to about $500/kilogram, certain kinds of businesses could justify using near-Earth orbits. If that happens, the existence of a backbone architecture that can support commercial uses in space may become essential the same way that the Internet is becoming an important infrastructure for commerce on Earth.
Dr. Cerf ended by commenting on the rapid pace of advances surrounding the Internet and the complexity of issues to be resolved as more and more of the world’s population comes online.
Dr. Spencer invited questions on Dr. Cerf’s talk. Dr. Wessner began by asking what might be the show-stopper in this otherwise rosy scenario of Internet evolution. Dr. Cerf replied that technological disaster, such as lack of transmission capacity, was a possibility. However, this seemed unlikely because recent tests indicated that capacity is growing rapidly. Today the network runs 10-gigabit-per-second fiber optic lines with a capacity of 40 gigabits per second. Engineers had demonstrated a near-term capacity potential of a trillion bits per second a few months earlier, and more recently a test of seven trillion bits per second was announced. “So I don’t think that fiber will be a show-stopper.”
Dr. Cerf noted that developing routers that are fast enough to transmit packets of data to the Internet, he said, present a more complex challenge and a potential bottleneck. So far, he said, the router vendors have been able to develop higher capacity switching systems each time they are needed.
The Need for a Legal Framework
A more serious show-stopper could be the continued lack of a legal framework to support electronic commerce on the Internet. The treatment of content
must be consistent and simple to avoid great disparities “at the edges” and between countries. Dr. Cerf cited the example of a looming court case in France against the giant Internet company Yahoo. A user had posted content on a Web site that included Nazi memorabilia, and it is illegal under French law to post such content in public. Dr. Cerf was scheduled to be an expert witness in the case, and one issue he had been asked to resolve is whether an access provider might be expected to detect that Internet material might reach a citizen of a certain country. His answer would have to be no, because a provider does not know for certain from addressing data where the packets are going. He warned that policy decisions made by different countries could create sufficient disparities to jeopardize global commerce.
Dr. Myers of Xerox Corporation returned to the issue of network reliability, recalling the electrical blackout more than three decades ago when most of the power grid of the Northeast shut down. He asked who would be responsible if the Internet analog of a blackout or brownout were to block companies from doing business by halting the flow of information.
Dr. Cerf answered that the Internet service providers have the responsibility of providing a reliable infrastructure, but that the issues of interconnecting segments of the Internet have not been resolved and that the business models for interconnection between Internet service providers are not fully formed. By one model, the company that connects to an existing company pays for the connection. This does not resolve who pays when two parties connect to each other. Yet another model says that two companies of about the same size that exchange the same amount of traffic will not charge each other for the connection between them. Until the economics of such connections are settled there may be some difficulty in developing fully robust systems.
Where Should Regulation and Taxation Occur?
Dr. Aho returned to the subject of regulation to ask whether regulation would most appropriately occur at the source, in the transition, or at the point of reception.
Dr. Cerf answered that the point of regulation would probably depend on what is being regulated. Content, he said, is more easily regulated at the source. Content regulation is complex when it threatens to become censorship. For content that is widely agreed to be illegal, such as child pornography, regulation at the source seems sensible.
On the other hand, applying taxation to the Internet can be much more complex. One would have to determine where the transaction takes place and the parties involved in the transaction. Are the parties located in the host computer
that contained the information, or where your personal computer happens to be? Dr. Cerf suggested that geophysical location might become less and less important. For example, an online purchase might be made from anywhere in the world, so that the transaction might more reasonably be said to occur at the billing address of the credit card holder. For a cable system, a more appropriate regulation point might be the address of the subscriber—the delivery point. Such ambiguities must be resolved for Internet commerce to thrive.
Dr. Brynjolfsson concluded the discussion by pointing out the very large projected size of e-business. He said that the projections of $3.5 trillion a year in the United States and about $6.7 trillion worldwide are realistic and that such projections have been revised upward each year. He cautioned that care must be taken when comparing business-to-business volume with national or global GDP because the same good and its components may be sold at many Web sites, but would be counted only once in GDP. In principle, while the projections for future business-to-business volume may be reasonably realistic, they need to be considered in context, and are not directly comparable.