Lee W. McKnight and Joseph P. Bailey, Editors
Internet Economics
MIT Press, 1997. ISBN 0-262-13336-9. Pp. 526.

Reviewed by Terry Rooker
<trooker@CapAccess.org>

As the Internet continues to grow and expand, people have tried to define, model, and understand it. In Internet Economics, the economists take their turn. This book developed out of a conference held at MIT in 1995. The focus of the articles is the Internet architectural models and their implication for pricing and pricing policies in general. The stated goal of the book is to develop a metric for economic analyses of Internet transactions. I'll talk more about that later.

There are six sections to the book: Introduction to Internet Economics, The Economics of the Internet, Interconnection and Multicast Economics, Usage-Sensitive Pricing, Internet Commerce, and Internet Economics and Policy. The number of articles in each section varies. The articles also vary in how much economic theory they contain. Only one or two have detailed derivation of formulas and equations, so most of the material is accessible to those without an economics background.

The technical detail doesn't require a degree in computer science. There is a brief description of protocols, enough to understand the development of the architecture models. Because economics involves optimizing resource allocation, it is important to understand the difference between some of the protocols. But the discussion doesn't get into the technical issues, such as increasing the size of the Internet address space.

Trying to find a good model of the Internet has some potential benefits. Presumably, the economists could use it to devise a better scheme of charging for Internet services, which presumably would mean users would pay for the resources they use. Unfortunately, most of the payment schemes mentioned in this book involve usage-based charges. It makes sense if you are trying to distribute the charges to those who actually used the service. The book generally discusses two options for charging for services: the usage-based system I just mentioned and a flat fee scheme that is what most current ISPs use. The problems with the flat fee system, from the perspective of developing a model, are that everyone pays for the infrastructure, whether they use it or not, and the difficulty of determining exactly what part of the infrastructure is justifiably included in the flat fee. Because this system doesn't optimize the charges, it is usually rejected by the authors or is used as the example of what we don't need. Interestingly, the book doesn't mention the market forces that would solve the two problems. Basically, the ISPs will continually adjust their prices until they can pay for everything and still increase their customer base. That is what has been happening in the two years since the conference was held.

Most everyone agrees that usage-based charges would be better. Most of the economists writing for this book feel that they would prefer a usage-based system and have some good arguments in support of that type of system. Most users of Internet services would probably prefer that. Personally, I wonder about the flat fee because I don't do a lot of bandwidth-intensive work through my ISP and occasionally have troubles calling into my ISP. Why should I pay for the time that someone else is downloading lots of graphics?

There are some serious problems with usage-based schemes, from both a technical and an economic perspective. Economically, the problems involve defining exactly what a resource is. Of all the equipment and skills required to maintain an ISP, exactly how much of each should be allocated to the different Internet services provided? This also involves technical issues. Many Internet services maintain a continuous connection, such as Telnet. Others have intermittent connections, at least from the ISP to the host site, such as the HTML World Wide Web protocol. Both services require a continuous connection from the user's computer (or terminal) to the ISP. Telnet also requires a continuous connection from the ISP to the host, and HTML will have numerous short connections. Even worse, the bandwidth required for Telnet is much less than that required for graphics images downloaded to a browser. It gets even more complicated when you factor in network infrastructure and new services.

There is a major technical issue that is described as part of the problem with implementing a usage-based system. Even if you decide on what resources to monitor, how do you monitor them? The necessary hardware and software would have to be designed, built, and installed. This would add to the expense of operating an ISP. In addition, this monitoring would be using resources itself. Plus it is not clear that all resources would be something that was easy to monitor.

The Internet has refused to be categorized or defined before. Apparently, it will add economists to the list of those who have discovered this difficulty. Despite the efforts of those who contributed to this book, they cannot resolve the difficulties of defining the terms to switch to usage-based pricing for Internet services. The one theme through the book is that although usage-based pricing is preferred, it is too difficult to define with our current understanding of the resources required. Until those issues get resolved, a flat fee system is preferred because it is easier to implement.

This is an interesting book. That it can be the subject of such detailed analysis is probably a sign that the Internet has "arrived." As a technical professional, I never looked at some aspects of an Internet connection in this way before. It certainly opened my eyes as to what an Internet connection may actually cost ­ beyond just calling up ISPs and asking for quotes. As new services are proposed, these issues may become even more important. With its continued growth, the Internet is reaching a point where some resources are saturated (some will even claim they are oversaturated). Usage-based charging may be necessary to relieve some of the saturation so it is important to begin discussing these issues. As a first step in that direction, I recommend reading this book.