A new paper by authors associated with Rensselaer Polytechnic Institute and the University of Nevada Reno concludes that net neutrality is a bad idea. This isn't surprising - not because it makes sense, but because it was largely underwritten by AT&T, which hopes to make billions of dollars from throttling the free flow of traffic on the Internet.The paper, Value of Supporting Class-of-Service in IP Backbones, was co-authored by Murat Yuksel, University of Nevada-Reno, Shivkumar Kalyanaraman, Rensselaer Polytechnic Institute, K. K. Ramakrishnan, AT&T Labs Research, and Joseph D. Houle and Rita Sadhvanim, AT&T.

The paper compares IP backbone networks providing a single class of service for all traffic with a differentiated service - a network that provides different levels of service for different applications. It does some impressive math to tell us what we all knew intuitively - that a truly neutral network requires more capacity to deliver the same volume of traffic as a "Diffserv" environment. The answer turns out to be 60 to 100 more capacity, in fact.

But that's not exactly news. And didn't the title of the paper say something about "Value"? Yes it did, and along comes David Isenberg, who has forgotten more about the Internet than most of us will ever know, to point out in his blog on Research on Costs of net Neutrality that the paper in effect compares apples and oranges in order to reach its conclusion.

The paper was first publicized in an item in Silicon Investor. Isenberg responded:

SI quotes my former Bell Labs colleague and paper co-author KK Ramakrishnan as saying, "Clearly, an undifferentiated network in this context is less efficient and more expensive . . . We believe understanding the real impacts of the alternative strategies is important as the debate about network architecture unfolds."

I think the paper's introductory assertion goes way beyond the scope of the study, and KK's quote, if indeed he is accurately quoted, goes way beyond the data presented.

(The paper doesn't say anything about the cost of implementing and maintaining Diffserv on a network versus the cost of increased capacity, either. Or the value of a classless network to its users, who might just find that they'd rather not delegate to AT&T what that network should be allowed to carry.)

The problem with the paper, according to Isenberg, is that it doesn't address the costs of implementing DiffServ versus more capacity. He asks, "Does 60 to 100 percent more spare capacity equal 60 to 100 percent more cost? Not in my experience. My 20 Mbit FIOS connection costs me about 20% more than my 768 kbit DSL connection."

He finds the failure of the authors to extend the conclusions from capacity to raw costs of capacity is deliberately misleading, especially, he says, when the researchers invoked "economic viability" and "cost of capacity" in their introduction to the work.

The Internet is a public utility - in fact, it is perhaps the greatest public utility in history - and it should be treated like one and operated for the good of the public. Companies like AT&T didn't complain about the rules when they got into the business of providing classless service to Internet customers. Just because they've figured out how to bend those rules to their advantage this late in the game is no reason for them to be permitted to do so.

If building and selling classless capacity has been a good business for the entire history of the Internet, what has suddenly made it a bad business? AT&T needs to answer that question honestly and directly, rather than serve up cheap propaganda covered in academic sauce.