The following text is copyright 1998 by Network World, permission is hearby given for reproduction, as long as attribution is given and this notice is included.

Developing an anti-Internet

By Scott Bradner

I'm starting to hear about an Internet I do not recognize.

A number of speakers at the recent Vortex98 meeting (referred to by John Gallant in his editorial in the May 25th issue of Network World) and some of the speakers at the Second International Harvard Conference on Internet and Society (http://cybercon98.harvard.edu) talked about an Internet that might have been brought to you by the old Bell Telephone System.

Oracle chairman and CEO Larry Ellison, speaking at the Harvard conference, described a Network Computer-based Internet in which the only thing that the average user would have is a web browser. In this Internet what is on the desktop is simple, very simple, and is supported by services in the network.

At Vortex98 three different speakers from companies that make big phone switches talked about their view of the future Internet where there has been some level of convergence between the current Internet and the current phone system.

The views of the Internet that these people described looks, on the surface, similar to the view that many of us have for the Internet of tomorrow. A ubiquitous connectivity service which supports various types of applications ranging from the current ever expanding web to real-time voice and video. But looking a bit closer one sees that their vision is of what might be called the anti-Internet.

The most important feature of the Internet is the ability to experiment. This ability comes from the use of common, open, standards-based interconnection protocols which are used to transport information for applications. Internet applications reside on the computers at the periphery of the Internet. These applications may make use of some support services scattered around the 'Net such as the Domain Name System (DNS) but can generally be run even in the absence of all services other than the network actually forwarding the packets.

This is different than in the phone network. In this case applications reside in servers that are operated by the phone companies as part of phone network. These servers are in the phone switches and in the service nodes. The user only has access to a very dumb node indeed, a telephone. New applications are added to the phone network by modifying the servers in the network, something that can not be done by the user.

I suppose we should expect to see people from the traditional telephony and mainframe worlds see the freedom to experiment that is the basic reason that we have the Internet we do today as confusing to the user. But their Internet is not an Internet that I would be all that happy in. Their Internet would result in the same dramatic lack of innovation that we have become all too familiar with in the phone system.

I'd rather the Internet that we have currently. Sure, some things could be better, controllable quality of service, for example, and we (the IETF) are working on that, but when you consider the alternative vision being presented, the risk of a little bit of confusion does not look all that terrible.

disclaimer: At Harvard confusion, for lack of a better term, is good, since it is the unconfused who have stopped thinking. But the above confusion is my own.