Weaning the Net
Amidst all the self-help chats, computer-game tip exchanges, militia meetings, flame wars, and furtive cyber-gropes, there echoed through the Internet a brief, hollow silence at midnight, April 30. Hollow because no one, from the elysian fields of the WELL to the airport mall of America Online, ever really heard it. There was nary a glitch in the brave new wired world as the NSFNet network -- once the main route, or "backbone," of the entire Internet -- was deactivated by its masters at the federal National Science Foundation (NSF). So ends an era as momentous as its passage was unnoticed. For a quarter-century, from its beginnings as the nuke-proof ARPANET defense network to its last seven years under NSF's $12 million yearly subsidized wing, the Internet has grown from an academic tool into a major cultural force and, not incidentally, a highly profitable business. And now it has made the leap from government-funded playground to corporate battleground. The NSFNet shutdown finishes an orderly two-year retreat that began when the NSF decided the Internet had grown big enough to take care of itself. By limiting commercial traffic on the net backbone for years, the NSF stimulated the growth of alternative private sector backbones; today there are at least seven, run by companies that include phone giants MCI and Sprint. In November of 1994, NSF advised the regional networks that comprise the greater Internet to disconnect from NSFNet and negotiate contracts with the commercial providers. By May, the transition was complete. Without a doubt, the Internet will hold together. But whether it will remain affordable to the average American is no longer a certainty. NSFNet service was never free, but its rates were fixed; with its departure, a big, wholly unregulated business has sprung up overnight. The new backbone players will have no limits on what they can charge; on the other hand, they must compete with each other for customers, setting the stage for net price wars that might ultimately benefit consumers. But James Love doesn't think so. He's the director of Ralph Nader's Taxpayer Assets Project, which recently received a Rockefeller grant to study Internet pricing. Love fears that the net's new masters will use the four Network Access Points -- the hubs of the Internet where all the new backbones meet -- as "choke holds" through which Internet service might be metered. In such a plan, the various backbone companies could institute a per-megabyte charge for data traffic on their networks. The Internet carriers who buy service from the backbone providers and resell it to individuals -- the WELLs, ECHOs, and Pipelines of the world -- might then be forced to raise prices. Colleges and research facilities, formerly NSFNet's biggest customers, might have to ration service. "There's not exactly going to be a vote on it when it happens," says Love, whose organization unsuccessfully pushed an amendment to the House telecommunications bill that would have required the FCC to hold a public online forum on Internet pricing. "As far as I'm concerned, if you look at the commercialization process, the telephone companies' self-restraint is the only thing keeping us from measured-use pricing," says Gordon Cook, publisher of the Cook Report on Internet-NREN newsletter. "I think nobody really knows until somebody really tries it. I would hope that market competition would succeed in preventing it, but it would be a hope and nothing more." Other experts are far more optimistic, including Anthony Rutkowski, the executive director of the Internet Society (ISOC), an umbrella group for the various industry and research committees that set the technical standards for the Internet's operation -- the net's "provisional government," as it were. Noting that many carriers switched to commercial providers months before the NSF cutoff, he claims that "the financial impact of the 'pulling out' is actually pretty minuscule -- and more than offset by the removal of commercial restrictions and the effecting of a massive competitive marketplace for access that has provided widespread connectivity at very minimal cost. So the bottom line is that the real cost to the vast majority of users has actually come down significantly." Case in point: While universities may have benefited most from the NSFNet, Harvard University had five different bidders when it contracted out its backbone service, and is now paying less than it did before, according to Scott Bradner, a consultant with Harvard's Office of Information Technology and an ISOC board member. But the competition may not be spread as evenly outside Cambridge, says Barbara O'Connor, chair of the Alliance for Public Technology. "In rural or geographically isolated markets, where there aren't major competitors, I'm not at all sanguine that rates will go down," she says. According to Jon Postel, a director of the Information Sciences Institute of the University of Southern California and another ISOC board member, metering the flow of data is, in most cases, an enormous technical problem that is prohibitively expensive. As an early designer of many of the Internet's key protocol specifications, including its most basic, TCP/IP, he ought to know. Still, both Cook and Love say they've heard reliable rumors that some net providers will institute measured service in the near future. MCI and Sprint would neither confirm nor deny the prospect. "We're not going to do anything that's going to harm or hurt the Internet," said MCI spokesman John Houser. "MCI is pro-Internet, and very pro on the Internet continuing to grow. If I were to look at a crystal ball, I would see a mixture of pricing plans." The infrastructure business is not the only thing the NSF is abandoning -- it's also getting out of the standards game. Over the next two years, it will cut all government funding for two of ISOC's most vital subgroups--the Internet Engineering Task Force (IETF) and the Internet Network Information Center (InterNIC). IETF sets and maintains all protocols and standards for the Internet; InterNIC assigns every block of e-mail addresses and ``domain names'' (what you see after the ``@'' sign in an e-mail address) in the U.S. For years, they have performed the minor miracle of managing the world's largest, most diverse communications network with staff sizes and budgets well below that of a New York City municipal agency. With the Internet doubling in size each year, the need for these groups will be more critical than ever. But with government funding pulling out, how IETF and InterNIC will fund their budgets (about $1 million and $3 million to $5 million, respectively) is ``the hot topic right now,'' says one ISOC board member, who notes that the decision may be made at ISOC's annual International Networking Conference, which was in progress at press time. The most popular solution is to get funding from industry; the InterNIC would sell blocks of addresses to service providers, and telcos and hardware companies would provide grants to run the IETF. ISOC director Rutkowski notes that InterNIC's counterparts in other countries are all already funded by service providers. ``The ideal way is to fund central assignment authorities via fees from the providers. That's the way the North American telephone authority works,'' says Bradner. As commonsensical as this solution may be, it has ``ignited controversy both within and outside'' the ISOC, according to the April 3 Network World. And in the case of the IETF, some critics also fear an ``agency capture'' scenario should major industrial players begin to fund the group. One or more companies, says Cook, might push the Internet standards in a direction that would benefit them. ``He who gives you your money for room and board can call the shots,'' says Cook. ``How on earth would we have any guarantee the standards process would remain open?'' Most experts say such an outcome is unlikely. ``I really don't think so,'' says Postel. ``It would be a mistake to allow one corporation to overwhelmingly sponsor (the IETF). But if each antes up an equal amount, then you would see a reasonably fair environment.'' Postel adds that while most of the IETF committees are already composed of industry employees, they remain ``independent individuals...looking for the best solutions.'' And, according to Rutkowski, industry funding of the IETF isn't even a foregone conclusion. ``The IETF Secretariat could probably be self-funding by simply raising the IETF fees from the $150 for food (at its triannual conferences) to about $350,'' he says. ``At current attendance levels, this would generate about $600,000 per year.'' When we think of hot political issues affecting the Internet, it's always the sexier ones that come most readily to mind, that most easily turn out the cyber-activists and fill up the net-bound petitions: stuff like electronic privacy, freedom of speech, protection from (and sometimes of) hackers. Unquestionably, these good fights must continue to be fought. But as these battles unfold, the players must be fully aware of their arena--the underpinnings of the net itself and how its changing contours circumscribe their struggles. (MCI certainly is; two years ago it hired ISOC president Vint Cerf to head up its Internet division.) These contours will do a lot more morphing once government involvement in the net infrastructure draws to a close. To call such changes ominous is clearly an overreaction, but to call them trivial is a mistake.