A map of the report, chapter by chapter

The Effect of Internet Use on the Nation's Telephone Network is organized like a legal brief rather than an essay. Knowing what each of its five chapters is for makes it much easier to find the part you came for.

What the report physically consists of

The title page carries four things: the authors, Lee L. Selwyn and Joseph W. Laszlo; the client, the Internet Access Coalition; the date, January 22, 1997; and the imprint of Economics and Technology, Inc. at One Washington Mall in Boston. Below that sits a reservation of rights. ETI holds the copyright and the notice forbids reproduction, so the chapters here are described and quoted from rather than reprinted.

The body runs past fifty pages and divides into an executive summary and five numbered chapters, supported by four tables and four figures. A separate Appendix A, analyzing second telephone lines attributable to online service use, was published as an Acrobat file of about 105 kilobytes and carries the arithmetic behind the report's revenue findings. Neither that appendix nor the original figure images survive in any capture, which leaves this contents page as one of the few records that they existed.

Why the argument runs in this order

The report does not build to its conclusions. It states them in chapter one and spends the remaining four chapters earning them, which is the convention in expert testimony. A regulator or a rival economist can see the claim before deciding how much of the supporting work to check.

The order of the support is deliberate. Chapter two establishes physical facts about switching equipment that both sides would have to accept. Chapter three uses them to attack the carriers' evidence as a class, on grounds of sampling and of omitted revenue. Chapter four then takes each filing in turn, so no company could say its study had merely been lumped in with the rest. Only then does chapter five turn to remedies. The sequence matters because the accounting critique depends on the engineering: a reader who does not accept that congestion occurs in specific, fixable parts of a switch will never accept that the problem is local rather than systemic.

"The imposition of per-minute access charges for the use of the current circuit-switched network is not the solution to any problem that may exist."

Chapter 1, The Effect of Internet Use on the Nation's Telephone Network

What each chapter does

Chapter one, the introduction, states the three propositions the rest of the report defends: that data traffic posed no significant threat to network integrity, that the carriers had overstated its costs while ignoring the revenues it generated, and that competition and data-friendly technology, not per-minute access charges, were the long-term answer. Chapter two, on network and switch architecture, is the technical foundation. It takes a typical Class 5 central office switch apart, component by component, to show where calls can actually be blocked, and notes that carrier tariffs charged a premium for the configurations that would have served high-use lines most efficiently. It also makes the off-peak point: online use clustered outside the voice busy hour, filling capacity that would otherwise have sat idle and pushing average per-minute cost down rather than up.

Chapter three, the general assessment of the carrier studies, does most of the damage. It argues that the filings rested on anecdotal evidence from a few worst-case central offices chosen precisely because their traffic was unusual, that internet users imposed no disproportionate cost, and that they had become a major source of new revenue. It distinguishes efficient from inefficient ways of connecting a provider to a switch and contends that the carriers had attributed all such traffic to the inefficient kind. It also notes that congestion inside the internet itself had been misread as congestion on the telephone network, and that the carriers were preparing to sell internet access of their own.

Chapter four drops the general case and goes company by company, assessing the Bell Atlantic, NYNEX, Pacific Telesis, U S West and Bellcore studies in turn and finding that each reached the same conclusion by leaving out the same facts. Chapter five is short. It restates that the filings did not justify access charges on internet and enhanced service providers, and points instead toward competition and data-friendly transmission technology.

The tables and figures carry the evidence

Four tables do the quantitative work. Table 1 compares what carriers charged for PRI ISDN, for digital trunk groups and for ordinary analog business lines, showing how pricing discouraged the efficient serving arrangements. Table 2 sets the sources of carrier costs and revenues from calls to enhanced service providers side by side. Tables 3 and 4 hold the second-line analysis: roughly six million residential lines nationally attributable principally to online access, and their cumulative revenue since 1990. Table 4 puts that cumulative figure at $3.6 billion, while the coalition's executive summary says more than $3.5 billion for 1990 through 1995, a small gap worth knowing about before quoting either number.

The four figures are diagrams. Figure 1 traces the path of a call from a subscriber to an online service across a local exchange network. Figures 2 and 3 open up a Nortel DMS-100 switch serving a provider, showing which internal components a modem call occupies. Figure 4 plots the growth of online services against demand for additional residential lines, the visual form of the revenue case. None of the images survive; the diagrams exist now only as their captions describe them.

Choosing a way into the report

For most readers the executive summary is enough. It compresses the argument into a few pages and is the version the coalition distributed itself. A reader interested in policy rather than plant should go from there to chapter three and then chapter five, which carry the economic case and the recommendation without any switching detail.

Readers who came for the technology should go to chapter two, the most detailed surviving public account of how a mid-1990s central office handled a modem call and where its real limits were. Anyone interested in how regulatory arguments are built will find chapter four the most instructive, since it shows five corporate research efforts arriving at a convenient conclusion and lays out how each got there. The Wayback Machine holds captures of every chapter at its original address.