jane@hpuxa.ircc.ohio-state.edu (Jane M. Fraser) (01/04/91)
[Moderator's Note: Jane M. Fraser wrote this article which will appear in print later this month; she has kindly provided an advance copy to TELECOM Digest for your consideration. PAT] [The following article will appear in the January CAST Calendar. To be added to the hard-copy mailing list for this newsletter reply to this message or write: Center for Advanced Study in Telecommunications, 210 Baker Systems, 1971 Neil Avenue, The Ohio State University, Columbus, OH 43210] The Internet is network of computer networks used primarily by educational and research establishments. The parts of the Internet that have been funded by federal resources (for example, NSFNET) may be used only for activities that support education and research. Other parts have not been so funded, and usage is not restricted. Various proposals have been made to extend the Internet to more institutions, to allow commercial use on all parts of the Internet, and to increase the bandwidth of the federally supported part of the network. On November 29 through December 1, I was one of approximately 150 attendees at a conference addressing various issues about the future of the Internet. I have always felt very confused about what is the Internet, what are the restrictions on usage, what different parts of the network are doing, and what options are open for the future. I learned one fact for certain at this conference: almost everyone else is confused also. I will report on some of the specifics of what happened at the conference, putting emphasis on aspects I think will be of most interest to the readers of the Calendar, but I am also confident that, no matter how careful I am, this report will contain errors. The conference, Information Infrastructure for the 1990s, was sponsored by two programs at the John F. Kennedy School of Government at Harvard University: Science, Technology and Public Policy and Strategic Computing and Telecommunications in the Public Sector. The two primary organizers were Lewis Branscomb and Jerry Mechling. The two-and-a-half days were heavily packed with presentations of commissioned papers, comments by panels of discussants, and open discussion from the floor. The main points the conference reinforced for me are, first, the growing importance of computer networks for fast communication and, second, the growing importance, for many users, of interconnectivity of networks. The first needs little comment. The second may be of importance more to some sectors, especially academics, than to others. Academics and researchers often want to communicate with a wide range of people and, thus, want to be able to send electronic mail to people on many different networks. Some companies may want their employees to communicate only within the company, not with those outside it, but others find interorganizational communication to be very important. Some networks already interconnect (although not completely), for example, AT&T Mail, CompuServe, and the Internet. Others are isolated, for example, Prodigy. Many barriers, institutional and technical, make it difficult to interconnect networks, but, I believe, there will be increasing demand from users to do so. At the federal level, a proposal has been put forth for federal funding of NREN, the National Research and Education Network, which would, roughly, be an extremely high bandwidth version of the Internet. (The latter sentence is undoubtedly not error free.) Most uses of supercomputers, almost by definition, require and generate huge amounts of data. For example, at the conference, we viewed a short tape of a simulation of the formation of a thundercloud. Remote access to supercomputers has always been cited as a justification for investing federal money in the Internet, and this again is one of the major reasons cited for the need for NREN. Indeed, the ability to create and manage a network at the data speeds being contemplated is itself viewed as a research issue. However, other participants argued that "low-end" use, that is, use not requiring high bandwidth, is also an appropriate topic for research. As the network expands and usage grows (which is happening at an amazing rate), questions arise about the ability of existing mechanisms to handle traffic. These participants argued that the networking of the large numbers of computers on the Internet (and its affiliates) is also worthy of attention, even without the addition of more bandwidth. This discussion of the importance of low-end use was naturally related to issues of allowing more general access to the Internet, for example, for K through 12 educational institutions. Currently, most academic users of the Internet receive access through their institution's connection. While the institution itself bears considerable cost, most academic end users do not receive a bill for usage. Internet connectivity to researchers is viewed by many academic institutions as being analogous to the library (for which usage fees are generally not charged to the end user or to the end user's academic unit), rather than analogous to the phone (for which such usage fees are charged). The user (or the academic unit) usually must provide a terminal or personal computer. Here at OSU, the computer magnus provides Internet access for anyone who requests it. (Actually, this is not quite accurate; magnus accounts will shortly be available to all OSU users.) One paper, "Pricing the NREN: The Efficient Subsidy," by Gerald Faulhaber, presented an economist's arguments against current pricing and subsidization schemes. Several commercial enterprises have been created (for example, PSI) to provide Internet access for commercial enterprises. Recall that commercial use is allowed as long as the use is in support of research and education. For example, a researcher at a commercial enterprise can communicate with researchers at academic institutions on research topics. A company can also communicate with researchers about its products. Two commercial users on different commercial networks must be very careful, however, since their communication with each other might traverse parts of the network on which commercial traffic is forbidden. However, it is often difficult for the user to predict what route a message will take. If all this seems arcane and unclear, it is. Many people (including Alison Brown of the Ohio Supercomputer Center) are working to make these aspects less arcane and more clear. One paper, "The Strategic Future of the Mid-Level Networks," by Paulette Mandelbaum and Richard Mandelbaum, explored various possible models for relationships between commercial and educational enterprises on the Internet. A portion of the conference had an Ohio focus. Jerry Mechling visited Ohio this summer and interviewed many people in order to write a case paper, which was presented and discussed at the conference, An Information Infrastructure Strategy for Ohio. Partly because of this, we had a fairly sizeable Ohio contingent at the conference: Gerald Anglin (Litel), Alison Brown (Ohio Supercomputer Center), Sally Cousino (Ohio Bell), Nick Farmer (Chemical Abstracts), myself (CAST), Jerry Hammett (State of Ohio), Don Olvey (OCLC), Tim Steiner (State of Ohio), and Ron Vidmar (State of Ohio). I found one of the most successful parts of the conference to be our caucuses, both before and after the conference. Other papers presented at the conference included "Information Infrastructure for the 1990s: A Public Policy Perspective," by Lewis Branscomb; "Technology Issues in the Design of the NREN," by Leonard Kleinrock; "Life after Internet: Making Room for New Applications," by Larry Smarr and Charles Catlett; "A Coming of Age: Design Issues in the Low-end Internet," by Ken Klingenstein; and "The NREN as Information Market: Dynamics of Public, Private, and Voluntary Publishing," by Brian Kahin. Copies of all the papers are available for loan from the CAST office. There were also smaller sessions involving presentations on current uses of the Internet. One presentation was by Allan Weis, from Advanced Network and Services, Inc., ANS, a "nonprofit organization dedicated to the advancement of education and research." ANS is funded by IBM and MCI to help build computer networks. As with all conferences, some of the most important discussions went on in the hallways and at meals and some of the most important results were the contacts made. Despite my dismay at finding myself at a conference with presenters who were all white males (including one who addressed the group as "gentlemen"), I think the conference was excellently organized and run. I applaud the organizers for focussing us on such an important issue: information infrastructure for the 1990s.