by Joe Flower


A version of this article is published in New Scientist, June 1995
Please see our free downloading policy.


  1. Ripping out the backbone
  2. Spawn of the military/industrial complex
  3. Internet facts of life
  4. So, who owns the Internet?
  5. Who runs the Internet?
  6. What's popping up on screen now?
  7. Who pays?
  8. Scenarios for the Internet Future
  9. Critical factors
  10. How it works

Ripping out the backbone

The Internet had a backbone transplant in April, and it didn't seem to bother it one bit. In a process that was completed on May 1, the U.S. National Science Foundation's NSFnet handed off its responsibility for the Internet's original "backbone" to private hands.

The moment marked another stage in the astonishing proliferation and evolution of a medium some consider the greatest shift in human communications since Gutenberg first locked down a rack of moveable type -- and a good moment to ask: What's going on here? What is the Internet turning into?

Spawn of the military/industrial complex

The Internet started simply enough. Its incubator was the U.S. military/industrial complex -- and that shaped its two exotic features: 1) A true network, more like lace than a wheel, it had no center, no "head" that could be cut off by a nuclear strike; 2) It allowed all sorts of different computers to talk to each other.

Started in the late `60s and developed throughout the `70s, the original ARPAnet (administered by the U.S. Defense Department's Advanced Projects Research Administration -- ARPA) consisted of a single "backbone," a high-speed network of powerful computers at key universities and research sites linked by dedicated telephone cables and powerful computer switches, tying together many regional networks. But it would accept non-commercial traffic from anywhere. Networks spread across the globe often found it quicker and easier to route their traffic through the U.S. backbone than to talk directly to each other.

Its growth has been phenomenal. Since 1988, when ARPA turned over its responsibility for Internet to the NSF, the net has approximately doubled in size every year. By the end of this year, an estimated 25 to 30 million people worldwide will be connected to the Internet, and over 50 million people will be able to access it from other networks through email. Much of this growth was in the United States, but starting in early 1994, the number of networks in developed countries began to multiply rapidly. By February of this year, over 90 nations had a direct connection to the "core" Internet. The "matrix" Internet, that network-of-all-networks that hooks the "core" with other webs such as BITNET, FidoNet, AppleLink, Minitel, and UUCP, reached into 168 countries. Many smaller local and regional backbones were growing by 20 percent or more per month.

Internet facts of life

Everybody's got a vision of the Internet and its future: it's an erotic playground, a research tool, the new broadcast medium, the world's biggest shopping mall, an anarchic frontier, the fad of the moment, the next level of human consciousness. If we hope to sort through these claims, we have to keep a few basic bits of reality in mind.

1) It's many-to-many, mostly. Unlike broadcast and print ("one-to-many" media), the Internet's "consumers" and its "producers" are the same people. Most of the content of the Internet -- email, MUD and MUSE simulation games, research data, Web pages, UseNet conversations -- is put there by its users. What would television be like if you could plug your camcorder into the set and show everyone what was happening at your house?

2) It's cheap and getting cheaper. Each separate piece of the puzzle that makes the Internet -- computer memory, computing power, high-bandwidth fiber optic and cable connectivity -- is dropping in price with breathtaking swiftness. To take just one example: a decade ago, a gigabyte hard drive (capable of storing a billion bytes of information on a disk) cost tens of thousands of dollars and resembled a stack of pizza platters. Last fall, a gigabyte drive the size of a pound of coffee was going for US$1400, retail. As of this writing, the store on the corner has it for US$347.

The just-replaced NSFnet backbone cost only about US$22 million per year, counting the equipment, the leased lines, and subsidies to regional nets -- less than a dollar per year for each Internet user -- and the total U.S. cost for the Internet has been estimated at only ten times that much. Nor does the individual user need the latest equipment to get online. In fact, if you are willing to do without the whiz-bang graphics, you can do it with the kind of outmoded-but-still-working equipment that people give away.

Professor Hal Varian of the University of Michigan, an expert on Internet economics, considers the costs "minimal." Anthony-Michael Rutkowski, executive director of the Internet Society, speaks of the cost of bandwidth "driving toward zero."

Moreover, the Internet has been built on a tradition of software that is either free or so cheap that it might as well be, sometimes written by individuals, sometimes developed by large net-related institutions. A series of useful tools, such as the Mosaic web-crawler, http (the communication language of the Web), html (the display language of the Web), Gopher (a way of putting files on the net), Archie and Veronica (search tools), and Eudora (a mail program), have come to life this way.

3) It's easy, and getting easier. Only two years ago, getting involved with the Internet meant learning a string of Unix commands. Today, cheap and powerful new software, such as Netscape, Mosaic, Hot Java, and their rivals, have given "net-surfing" point-and-click simplicity. IBM's OS/2 Warp already provides "one button" access to the Internet, and Microsoft's Windows `95 promises to do the same.

4) It's got time on its side. Younger generations of decision-makers are increasingly comfortable with it, come to expect it, rely on it, and think of new uses for it. Increasing numbers of colleges and universities, especially in the United States, provide direct high-capacity Internet connections (T-1 or ISDN lines) in every dorm room, and much of the Internet is being built and maintained by the volunteer or low-paid efforts of graduate students and young corporate techies.

5) Finally, and perhaps most important, it's not under central control, not even at the center.

So, who owns the Internet?

No one.

The Internet seems a great mystery to many people. Some assume that the U.S. government, which sponsored its birth, must own it. Others assume it's a private company, like Compuserve or Prodigy.

But no one organization, company, or government owns the Internet. The Internet consists of the connections between millions of computers with millions of owners in over 90 countries. The NSFnet handed off its responsibility not to a single private company, but to a network of private U.S. backbone providers. This network is informally "kluged together," according to Varian: "They still have to work the bugs out." The routers that do the system's heavy lifting are owned by major data transfer companies and consortia, such as ANS (owned by America OnLine), PSInet, SprintLink, NETCOM, BBN, UUNET, Merit Network, MCI, Pacific Bell and, in Europe, the European backbone ("EBone"), Eunet, and Dante. Some eighty percent of the net's traffic is carried by the three largest providers: MCI, ANS, and Sprint.

These companies make their money by charging access fees to the Internet providers, who in turn charge access fees to the institutions and individuals who want to plug in. Nobody owns the whole system.

The U.S. federal government has subsidized the development of many basic pieces of the Internet through various agencies. The National Science Foundation's subsidies of regional networks will step down through the end of the decade. The NSF and other agencies have also paid for a lot of the scutwork such as administering Internet address assignments, and the Internet Society's standards work, and paid for the first four network access points (NAPs) for the new, non-government version of the "backbone."

Who runs the Internet?

Nobody, in a sense. There is no czar of the Internet who could turn it off tomorrow, no Internet Senate that could send it careening off in a new direction by taking a vote.

The basic work that keeps the Internet running and growing is done by the Internet Society, a series of committees, working groups, and task forces, made up of technical people from the major users and vendors of the Internet (government agencies, academic institutions, and major service providers). These groups design the technical standards that allow the computers to talk to one another. Much as a household appliance has to use the standard voltage if you want it to work when you plug it into the local power grid, so any system that wants to hook up to the Internet has to work with the technical standards set by these groups.

"The Society has no coercive power," according to founder/president Vinton Cerf, a senior vice president at MCI and one of the original architects of the Internet. "Nothing they do is enforceable. It's all enlightened self-interest. The real secret behind the Internet is that it's a grass-roots, bottom-up system."

The central standards body, the Internet Engineering Task Force (IETF) works constantly online, and meets face to face three times a year. According to Anthony Rutkowski, executive director of the Internet Society, these meetings "bring together more than 500 people at the meeting location, with multicasting to more than 600 additional sites around the world. More than just developing standards, the IETF actually is a sophisticated technology transfer engine in which creative developers in academic, research, and business environments are joined in a kind of robust creative `soup' in which they imagine, write code, criticize, test, and very rapidly scale new information tools and services, free from stifling formalities and positions."

Rapid "scaling" to accommodate amazing growth is at the root of two of the biggest problems the Internet Society is wrestling with at the moment, both having to do with addresses. Internet addresses for local networks (such as "feat.newsci.ipc.co.uk") actually represent 32-bit numerical addresses, much like telephone numbers. But unlike telephone numbers, Internet addresses are not hierarchical. If you dial "011" on any telephone in the United States, the system knows you are making an international call. If the next two digits are "44," you're headed for the U.K. If "171" follows, you've got London. Internet addresses are random -- every major Internet router has to carry an enormous "look-up" table with the whereabouts of all 50,000 Internet-addressable networks, and each of these tables must be updated constantly. The Society is working to change to a hierarchical numbering scheme, which means changing everyone's address, without interrupting service.

And 32 bits will soon not be enough. "My guess," says Cerf, "is that today's 50,000 networks could turn into 1.5 million by 2000, and 1.5 billion by 2010, when a network may be located in something you wear, like a watch, or in your house or car." The Society is working on a standardized 128-bit address. "That's enough numbers," says Cerf, "to give a unique address to every electron in the universe."

The fact that the Internet has no center, and is so cheap, means that it is not dependent on national governments or large institutions -- yet it is so useful that many large institutions are willing to support it, use it, and design software for it. This makes it far more robust and hard to kill than any centrally-supported organization.

What's popping up on screen now?

The World Wide Web has grown like an arachnophobe's nightmare. This protocol allows people with "web browser" software to receive information in graphic form (text, tables, charts, pictures, and even audio and video clips) from tens of thousands of sites around the world. Virtually unheard of two years ago, it had grown to 14 percent of the total data on the Internet, surpassing all other forms of traffic, by the beginning of 1995. "I believe the Web will represent one-half of all Internet traffic [by the beginning of 1996] because it is the easiest and preferred way of presenting information, either for free or for sale," says Tim Krauskopf, vice president of research and development at Spyglass, the company that has developed the Enhanced Mosaic web browser. One key part of that thought is "for sale." Until a few years ago, commercial speech was banned on the Internet. Now it proliferates -- at a growing level of sophistication. Last year, Lawrence Canter and Martha S. Siegel, authors of How to Make a Fortune on the Internet, demonstrated their technique (called "spamming" by Internet denizens) by posting a commercial message to thousands of UseNet discussions, almost none of which had anything to do with what they were advertising. Internet users reacted angrily -- and in clever "hacker" fashion as well, inventing "cancelbot" programs that can detect such repetitive messages and cancel them wherever they appear.

Such crude, invasive techniques are rapidly giving way to marketing that is far smoother, more selective, inviting, and even fun. For instance: Imagine a free email service. Dial a toll-free number from anywhere, log on, and send and receive all the messages you want. It won't cost you a dime. The only catch: the email shows up on your screens in envelopes, each with a "stamp" that's a corporate logo. Click on the envelope, and you get your message. Click on the logo to find out more about the advertiser, enter contests, or get coupons you can print out and use. No fantasy -- ProductView Interactive of Cambridge, Massachusetts has already announced the product, and advertisers are said to be lining up.

But the business uses of the net do not amount to just an ad here and there, or a sponsor's logo at the bottom of the screen -- they stand likely to transform the net, and the net in turn is likely to transform business. The Internet Shopping Network has been signing up 400 people a day since it opened its doors April 1. Over 60 organizations have grouped together in a consortium called CommerceNet formed to make it easier to use the Internet to do business. New net browsers include "unbreakable" encryption functions that will allow people to use their credit card numbers online with confidence, and let businesses connect up their companies' internal networks over the Internet in full privacy. MIT, General Magic, and other organizations are writing software "agents" that, among other things, will have the intelligence and legal authority to make purchases for you, without checking home, on the net. Other software, such as the new SurfWatch, will make parents more comfortable with allowing their children to shop and browse and shop in cyberspace, by blocking them from logging onto adult chat channels and other X-rated net sites. For good or ill, some net providers and advertisers are working up software that can log users' "clickstreams," the pattern of their online interests, so that they can target them with offers that they won't want to refuse. Many phone companies and such private firms as The McKinley Group are producing online Internet Yellow Pages to match you with the product you need.

At a recent conference, Sun Microsystems Chairman Scott McNealy declared, "Within the next few years, business-to-consumer electronic commerce could begin to replace much of the world's existing business infrastructure."

Today's email, Usenet conversations, chat rooms, Web sites, and shopping malls are probably only the beginning of an extraordinary proliferation of uses. Already software exists that allows users to send faxes, and even primitive voice telephone calls, over the Internet. AT&T's Vistium Share software allows people to collaborate on the same file at the same time over the net. The newest browsers, such as Sun MicroSystems' HotJava, allow people to build audio, video, and animation into Web sites, to hold live online auctions, and scroll instantaneous scores across the screens of sports fanatics -- and these uses will grow as the Net's capacity grows. The Internet's packet-switching technology is capable of carrying everything from voice to video, and is so much more efficient than other methods, the Internet may evolve into the great "integrated services network" of the future, carrying everything from voice telephone calls, paging, and personal communications services to cable television.

Who pays?

The Net has grown on a tradition of charging flat rates. Popular commercial systems like Compuserve charge by the minute, but on the Net itself, you pay for the capabilities of your connection -- the size of the "pipe" -- not for how much you use it. That's one reason it's so popular. It costs no more to send a note to Auckland or Oakland than to send it across town. The various providers have a "zero settlements" policy -- they assume that the service they provide others is roughly equal to the service that they receive.

But obviously some uses burden the system more than others. Videoconferencing, downloading Jurassic Park or a major piece of software, working interactively on the same piece of graphics from a thousand miles apart, or exploring a three-dimensional "virtual world," all use a lot more of the system's resources than sending a note to a friend or logging onto a chat service. A piece of video email might take 10,000 times as many bytes as regular email carrying the same words.

Most of the new uses of the Internet are hogs for bandwidth and memory. Cerf concedes, "We are worried about interactive video and voice requirements." Every piece of the system must upgrade its capacity rapidly and constantly to provide for these new uses. The switches, for instance, the "backbone links," will have to grow from T-1s, which can carry 1.5 million bits per second (Mbps) to T-3s, at 45 Mbps, which carry most of today's backbone traffic, (imagine a 20-volume encyclopedia in half a minute), or even OC-3s, at 155 Mbps -- which will carry most backbone traffic in six months to a year.

Who will pay for these? If the Net keeps to its "flat-fee" tradition, everyone will. It is possible that the price of these upgrades will fall so rapidly, and the number of new users willing to pay a monthly fee will rise so fast, that the money will be there. It's possible that those monthly fees will have to rise to pay for the new capacity. But there's a third possibility.

Net software honchos are building ever newer versions of the Internet's "operating system," TCP/IP. The newest version, IP6, carries a "flow ID" in the header to identify interactive uses for routing purposes. The same ability could be used for billing purposes, to charge heavy users more for the heavy use, or even to allow people who want special treatment, head-of-the-line status during peak hours, to get it and pay for it -- and this money could pay for the upgrades needed to carry all these wild new uses, without burdening the ordinary user who uses email and starts arguments on UseNet.

Telephone companies traditionally charge not just for the connection, but by time and distance -- through their lens, the net's flat rates seem crazy. And telephone companies, both the "Baby Bells" or RBOCS (Regional Bell Operating Companies) and the long-distance companies like AT&T, MCI, and Sprint, are getting more and more involved in providing direct access to the Internet.

But the problem is incredibly complex. For instance, in today's free-wheeling Internet, I might ask a question of a computer across town. The answer will come back in packets that are "from" that computer and "to" me. How would the net software know that it should bill the answer to me, and not to the computer that I queried? And the computer that I queried might assemble the answers from computers in Switzerland or Australia, or cause a cascade of actions on other computers that I don't even know about. It's a core value of the Internet that distance makes no difference. It's the ultimate globe-shrinker. And spreading all these resources around in hundreds or thousands of computers, which makes billing more difficult, instead of putting them all on massive central databases, where billing would be easy, is also one of the factors that makes the Internet so remarkably cheap, efficient, and interactive.

Charging extra for high use might be a good idea, might not. But who decides? Since there is no Senate, no Net Czar, what happens if one carrier decides to add on such fees, and others don't agree to pay those fees or pass them on? As Christy Hudgins-Bonafield, an editor for Network Computing puts it, "There is no Internet police force." The Net could Balkanize into local, regional, and proprietary nets that refuse to talk to each other.

Scenarios for the Internet Future

Over the next few years, the Internet will change just as rapidly as it grows -- but in what direction?

Scenario 1: Balkan ruins -- Things fall apart. Disputes over settlements and different methods of pricing split the Internet into systems that don't talk to each other.

"That's the hot issue for the next few years: the technology and economics of the interconnections between the different backbones and regional nets," says Varian, "The fear is that the net will balkanize. But I don't think it will come to that, since so many people have come to depend on it. There is such an economic advantage to being able to talk to each other, it's hard to see how the competitors could leave that money on the table."

"This seems unlikely," says the Internet Society's Rutkowski. "The greatest value of the public Internet is its connectivity. Virtually everyone understands that." A spokesman for MCI, one of the largest providers, said, "We are dedicated to the survival of the Internet. We would never do anything to damage it." By early June, most of the private backbone and regional systems had signed "peer agreements" that, for the moment at least, preserved the zero settlements policy.

Scenario 2: Hacker's Dream -- Price is not an issue, because the price of everything falls so rapidly and the efficiencies of the architecture are so great that the Internet in effect becomes nearly free. Large commercial enterprises keep to their own "cyber malls," leaving the rest of cyberspace unpaved. A free-floating "cybercash" economy develops which nowhere intersects the earth-bound banking system. The Internet evolves as a self-organizing, smoothly-functioning anarchy.

Rutkowski calls this "an extreme that has elements of the plausible."

Scenario 3: Yet Another Broadcast Medium -- The Internet becomes a cyberspace version of the Mall of America, dominated by big companies. Most of the content of the Net is commercial, is produced by professionals, and carries a charge, or advertising, or both. The uses of the Internet become so expensive that free or nearly-free use by individuals almost ceases to exist.

David Wetherell, CEO of Booklink Technologies in Wilmington, Massachusetts, a company that develops software for navigating the Internet, says, "My fear is that it becomes entirely privatized. The great thing about the Internet today is that no one owns it. If you have a good product or service, you can put it up on the network and let the public decide if it's good"

Donna Hoffman, a professor at Vanderbilt University who specializes in marketing in the new media, says, "For generations, the way to make money has been by reaching mass markets with a uniform message. A lot of companies are approaching the Internet with the same mindset. A slick commercial mass medium is now being developed, and it's very much in the one-to-many model." Hoffman and other observers fear that Microsoft and other major companies will attempt to dominate the Internet, effectively building toll booths up and down the information highway, and turn it into yet another broadcast medium. "But the mass market is dissolving," Hoffman says, "into a disjointed market of differing tastes. There are fewer and fewer things you can sell on TV. If you approach the Internet as just another mass market, you won't get anywhere. It is a new medium with a great potential for relationship marketing. The way to make money on the Internet is to take advantage of the medium, to allow people to customize their own experience. It's an opportunity to get closer to the customer than ever before."

Varian comments, "If we can solve the problems of coordinating multiple competing networks, we'll have a viable, competitive, low-priced industry. If we can't, there will be a huge advantage to being big, and the biggest guy will win. We'll have a monopoly or oligopoly, and prices will rise."

Vancouver science writer Barry Shell, though a critic of corporate intrusion on the net, is not too worried: "The Internet, with its open, distributed structure, was designed to withstand a nuclear attack. If it can do that, it can withstand corporate America."

Scenario 4: Unlimited bandwidth -- There is no "tragedy of the commons" in cyberspace, because you can always make more space. The "Mall of America" can coexist with the dreams of hackers.

As Rutkowski describes this scenario: "Diversity -- much like the real world, only much more accessible."

"This is what we are moving toward," says Hoffman. According to Varian, "If you build a system robust enough to handle interactive video, everything else is free, essentially. All the more traditional applications, such as email and file transfer, can just go along for the ride, because their demands for bandwidth are so much smaller."

Critical factors

There are many choke-points on the future of the Internet. For instance, the U.S. National Security Agency has for years been battling to suppress the use of so-called unbreakable public-key cryptography online, and a grand jury in California's Silicon Valley is expected soon to bring a federal criminal indictment against cryptography pioneer Phil Zimmerman for allowing his PGP (Pretty Good Privacy) cryptography program to escape onto the Internet. The goal of the NSA and some other elements of the U.S. government (and a number of other governments) is to force all net cryptography programs to have "trapdoors" accessible by law enforcement agents. Yet such trapdoors will render the software hackable -- and will greatly reduce the willingness of business and commercial users to trust it with commercial secrets, credit-card numbers, and financial transfers.

Other governments fear the Internet's freedom. China, for instance, has set an artificially high price in Internet access, and totalitarian governments around the world have set various restrictions on its use. Yet because of its highly distributed nature, restrictions and regulations on the Internet are fundamentally difficult to enforce -- and the net's role as a potentially powerful engine of economic development means that these governments cannot afford to simply ban it.

Ironically, what will most limit the growth and usefulness of the Internet of the future is its very power. A technology that can bring you all information from everywhere is useless unless it can help you discover just the information that you want, the people that you want to talk to, the conversation that you want to join. So a crucial test of the viability of the Internet will be the development of methods to filter all that information: software agents, mail handlers, and human intermediaries, such as net searchers, conference moderators and digest editors. In fact, a powerful point of view is becoming a valuable commodity: what does the world of the Internet look like through Rush Limbaugh's eyes? Or Bryan Eno's? Or Nelson Mandela's? It will be possible to make a living as a "host" or "lens" on the Internet, not by selling your own writings, but by selling your own point of view about what is interesting and important.

Intelligent software filters will become equally important. For instance, Oracle's "TextServer3 with ConText" claims the ability to search massive computer documents not just for key words, but for "meaning," retrieving relevant bits of text even when they don't mention the key words, and discarding bits with the right words but the wrong meaning. Similarly, CYCORPS, in Austin, Texas, expects to have software on the market by the end of this year that will have the ability to filter the masses of unstructured data on the net for just the information you need.

If it struggles past the choke points of excess regulation, and excess info flooding, the Internet will be a powerful force in the twenty-first century. "The Internet is much more than just a new kind of network for transporting data," says the Internet Society's Rutkowski. He speaks of it as "a fundamental transformation," an entire information infrastructure built from the bottom up, a "robust global mesh" of computers that allows "open collaboration in the hyper development and evolution of new technologies" and that will "transform the structure, methods, and individual skills within enterprises, institutions, and professions of all kinds. . . . A hundred years from now, history may well record the emergence and implementation of [the] Internet protocol as a profound turning point in the evolution of human communication -- of much greater significance than the creation of the printing press."

Technology | Articles | Main Page