The following text is copyright 1994 by Network World, permission is hearby given for reproduction, as long as attribution is given and this notice is included.

Paying the piper

By: Scott Bradner

I have said before on the pages that the Internet, despite the assertions of some people, is not a free good, but for most users it might as well be. Unlike telephone usage, most Internet service has flat rate pricing. An organization figures out what speed and type of Internet connection it's usage warrants and contracts for that service. The organization get a regular bill which is based on this choice and is independent of the actual usage of the connection. Most organizations then internally allocate the cost much like they allocate the cost of electricity. The end user never sees a bill and generally remains blissfully ignorant of the actual cost to the organization. (Considering the intellectual content of some of the mail on some of the mailing lists, one could end the previous sentence after the word "ignorant ".)

Flat-rate pricing, while prevalent, is not universal. There are a number of Internet providers offering one type or another of usage sensitive billing. For example the customer can get a T1 (1.5M bits/second) link and get billed a service fee quite close to the cost of a 56K bits/second link. The customer does pay a higher cost for the T1 phone line but the reduced service fee can be significant. The customer can then get the level of performance that a T1 connection provides at a substantial discount. The reduced fee remains in place unless the customer uses more than some set traffic threshold over some defined time span, in which case the fee is raised for that period.

Currently, in the U.S. all of the major national and regional network providers offer flat-rate service. This may not last. More and more of the major telecommunications providers are getting into the Internet service business. They bring with them a long history of usage-based charging for their telephone services. This will increase the number and variety of usage-based service offerings.

Some of you might wonder why an organization would want to switch from a flat-rate service to a usage-sensitive one, particularly in face of the vast expansion of World Wide Web usage and its heavy traffic demands. Well it would depend on the cost model. The cheaper T1 service described above is quite popular with the customers of some of the current Internet providers. It is a usage-based fee with very few (usually two) plateaus. There are many cases where a properly priced usage-based service could be quite attractive to an organization. They get cheap high speed access, assuming that they can control their appetite.

There are many potential effects of the introduction of this type of pricing on the current Internet culture. For example what would the effect be on the providers of all of that data in the anonymous ftp sites? The idea of paying extra to do the world a favor might not go down all that well. Another effect might be that organizations might start to do some kind of internal usage-based billing as they do for the phone message units and long distance charges.

The somewhat restrained joy at the prospect of these kind of effects brings one to wonder about finding alternate funding models for Internet (or actually, general data network) connectivity. The choice of funding paradigms for these services may significantly effect the type of Internet (or National Information Infrastructure if you will) we will be using in the future. Some of the pictures are less attractive than others. Paradigms that inhibit the prevalence of the available Internet resources are somewhat like building a housing development in the wilderness and wondering where the birds went.

In looking at the Boston Globe for Thanksgiving Day (November 24, 1994) I did a quick check and found that there were 54.7 pages of actual content, 107.3 pages of adds and over 300 pages of advertising inserts. Extending this model to the Internet would only please two Arizona lawyers but it might be a good idea to keep a careful watch on which forks in the info hypeway we start down.

Disclaimer: Internal allocation of the costs of data networking is an on-going issue at Harvard but the above wanderings are my own.