Since 1984, the National Science Foundation has funded what has historically been the highest-speed link in the network of networks which is the Internet. The NSFnet backbone was originally constructed to link the five NSF National Supercomputing Centers to each other. As researcher demand increased, so too did the number of research centers directly attached to the NSFnet. It was the need to limit the torrent of traffic on this taxpayer-funded backbone that led to the formulation of the NSF Acceptable Use Policy (which, among other things, forbade commercial use of the NSFnet infrastructure.) From that restriction sprang the commonplace (and entirely mistaken) belief that commercial traffic of any kind is forbidden on the Internet. Every two years since 1984, almost like clockwork, NSFnet has upgraded the bandwidth of its backbone from 56KB to T1 to T3 to the latest high-speed switching technologies.

Beginning this June and continuing for the next four years, the NSF is scheduled to progressively defund the NSFnet backbone. In its place, regional and larger commercial Network Service Providers (NSPs) will take over responsibility for providing access to backbone services for the Net.

Meantime, the NSF itself will fund the Very High Speed Backbone (vBNS), a technological successor to NSFnet. The vBNS will connect the original 5 National Supercomputing Centers (in San Diego, CA, Boulder, CO, Champaign, IL, Pittsburgh, PA and Ithaca, NY) with four Network Access Points (NAPs), which, theoretically, will be the only points at which the NSPs' own backbone services will tap into the vBNS. This is a VERY important transition in the evolution of the Internet. Acceptable Use constraints will mostly disappear, since the new technical architecture of the Internet will moot the issue of commercial traffic on publicly-funded hardware. The increasing commercialization of the Internet has already wreaked major changes in Net culture--and the physical rearchitecture of the Internet will accelerate this trend.

Then there's the Microsoft Network. At this writing, the Internet claims somewhere between 20 and 40 million users (the numbers are slippery because many of these 'users' are limited to email-only or email-and-Usenet-only access.) If Windows95 is a commercial success of the same order as Windows 3.x, it will boast something like 30 million users worldwide within the first two years after its release. Every one of those users will have an initial trial subscription to the nascent Microsoft Network (it will be packaged with the front-end software and both will be included in Windows95 itself.) That's an instant user community of roughly the same size as the Internet. (For comparison, let's note that CompuServe, America Online and Prodigy combined have fewer than 10 million users.)