"The Privatized NREN" Mitchell Kapor Electronic Frontier Foundation February 14, 1991 A Note on Terminology: Use of terms in discussions on networking is notoriously subject to confusion. I have chosen here to refer to the Internet as the current networkof networks connected by the NSFNET backbone. Some are now referring to this as the interim NREN. I have no quarrel with this usage, but will not adopt it here. I am using a term of my own coining, the national public network (NPN), to refer to the (still hypothetical) convergence of the NREN, the analog telephony public switched network (PSN) and its narrowband and broadband digital successor(s), the cable television distribution network, etc. Author's Note: In the small amount of space which the call for this paper required it is not possible to provide the necessary background to introduce and explain the context of various of the key stakeholders, concepts, and technical vocabulary employed. Those readers seeking further elucidation are encouraged to contact the author directly at the address supplied below. It should also be noted that the opinions expressed herein are the author's personal ones. Organizational affiliation is provided for purposes of identification only. Recommendation #1 The time has come to facilitate the transition of the Internet into the first phase of a national public network (NPN) by enabling a graceful transition to control and operation by the private sector. One of the successful outcomes of the Internet is that wide-area networking based on TCP/IP protocols has evolved from a research prototype to a level of maturity in which, if hardly ultimate, is sufficiently developed and robust to stand on its own. As NSF and other government agencies increasingly turn to new research on high-speed networking, the time has come to move current infrastructure into the private sector. Individual and institutional users, whether for-profit or non-profit, will benefit from decreasing costs and increasing levels of service through the dynamics of open competition in the marketplace. At the same time, the lapsing of usage restrictions will encourage the development of new varieties of commercial information and communication services which are offered over the network. Network access is becoming a commodity which should be purchased like any other computer or telecommunications service. The role of NSF or other government agencies with respect to providing network access should be provided on the same basis as they provide support for other types of computer equipment and services. This said, there are delicate questions as to how this transition is to be accomplished. Recommendation #2: Insure a level playing field for commercial, not for profit, and non-profit TCP/IP internetworking companies and institutions. The infrastructure should be one in which open competition is encouraged. A critical question which will determine whether there will be a competitive market for TCP/IP internetworking is whether and under what conditions will it be possible for an internetworking carrier to connect to the network. Will a single private party such as ANS effectively control access to the network through control of the backbone? To the extent that backbone access is required to connect to networks of other countries or to federally controlled networks as well as to mid-level networks, this is an even more serious matter. If so, and if the party has no obligation, legal or contractual, to provide interconnection, they could use this advantage as a competitive weapon to stifle the development of other carriers. This would be undesirable. The Internet, like other networks such as the voice telephone network, derives value from the universality of its reach. Any user within its universe may readily communicate with any other user. If a situation arose in which sub-communities of users were threatened with isolation from the rest of the net simply because their mid-level carrier (whether a regional non-profit cooperative, or national profit-seeking entity) was being arbitrarily denied access to the rest of the net, it would be an abuse of the public interest by the party exercising this manipulative power. A contractual obligation might be one which the NSF imposed in a further agreement between it and the party to cover the period subsequent to the expiration of the present NSF-Merit-ANS agreements. A legal obligation might be one imposed by a government agency such as the FCC to require interconnection. A model for this could be drawn from the rules for non-structural safeguards called for by the FCC in its Computer Inquiry III. It would be desirable to achieve the same ends as mandated by CI III's Open Network Architecture (ONA) without involving the constant, costly government involvement which mediates between the entrenched interests of monopoly owners of transmission facilities on the one hand and enhanced service providers on the other. It is my belief that the NSF has, in this critical transition period, a great deal of leverage on all parties to secure some form of voluntary agreements to these ends which would obviate the need to structure a highly-regulated TCP/IP internetworking industry, which no one really wants to do. These agreements should be committed in writing and made available to the public in order to ensure accountability. For instance, it might be possible for MERIT/ANS, as a key stakeholder, to voluntarily undertake some form of binding commitment which guaranteed other parties the right to interconnect on an equitable basis. Note that the situation under discussion is not that of the right of a node to connect to a carrier, but the obligation of carriers to provide equitable interconnection to other carriers. This parallels the rights of long distance telephone carriers such as MCI to connect to local exchange carriers. The author understands that the implementation of such a framework raises many large technical and policy issues which would need to be undertaken in order to make an open interconnection scheme work. For instance, there must be determined which services, in addition to basic IP transport, would form the "basket" of basic services which were standard to the entire infrastructure. Certainly naming services, but also emerging user directory services, information provider services, accounting services, and other as yet undefined services will need to be developed in a cooperative fashion. Recommendation #3: Internetworking carriers should adopt a usage policies which explicitly provide for non-interference with respect to the contents of user traffic carried through the basic transport services. Carriers should also be understood to have no liability for the content of these transmission. This mirrors the position of the telephone companies and other common carriers with respect to message content in those media. Note that other standards of care and liability, hence other usage policies, may be called for in the provision of enhanced services such as electronic mail, computer conferencing, etc. Unfortunately, space does not permit a discussion of these important issues here. Adoption of this recommendation would be most consistent with the first amendment right of free speech and freedom of expression. The policy mechanism by which this is to be achieved is not clear. It may be that the common law would support such a stance, but this is something which would only be known as the consequence of litigation. While it may not be necessary to take any legal actions in advance to achieve this goal, it is likely that there will be an atmosphere of uncertainty as to whether the announced non-interference with content policy will be upheld in the long term. I turn now to the issue of policies for the long-term NREN. Recommendation #4: Encourage information entrepreneurship through creation of NPN as an open architecture platform with low barriers to entry for information providers. There are important lessons to be learned from the rapid success of the personal computer software industry. In the PC world, applications developed as separate stratum from operating systems (the platform layer). Apple and IBM enabled growth of huge markets like spreadsheets and word processors by creating open architectures which encouraged third parties to risk their own capital and put ingenuity to the test by developing applications. The abstention of IBM and Apple from competing with VisiCalc and Lotus 1-2-3 was a necessary factor in enabling the growth of application markets. In the world of wide area networking and telecommunications there is an opportunity to transplant these ideas with an expectation of equal success as well, but it will require a bold new style of thinking and risk- taking on the part of the existing stakeholders. We must regard the NPN infrastructure as a platform. This implies that platform providers should not try to pre-empt competition by providing services, but should create open architectures which encourage the entry of new parties to create the applications. This does not mean that platform providers such as the regional operating companies should be denied the opportunity to participate, but it does mean that they should free themselves from the burden of assuming they will have to develop the enabling applications for this new platform. The key enabling applications for the new medium cannot be predicted in advance. Let the market drive innovation by making experiments cheap and lowering barriers to entry for providers. Competition provides rapid sorting process as successful applications and services are rapidly emulated and improved upon. The founders of Apple Computer did not anticipate the spreadsheet. They created a platform in which 10,000 new product ideas got a hearing in the market. Out of this, winners emerged naturally and swiftly. A good platform will encourage a large number of start-up organizations to take the risk themselves of developing a sustainable application or service, but only if the platform is accessible to them and if it is capable of reaching a large number of potential users on a commercial basis. Today we understand the immense popularity with "early adopters" of applications like wide-area electronic mail, computer conferencing, and electronic publications on the existing infrastructure. Yet these applications are caught in a peculiar limbo. The software is barely good enough for a technically astute person to understand and use. For the most part, users are not paying directly for these services. At the same time the commercial opportunity to further develop these applications is not widely perceived as so great as to cause firms to be willing to invest heavily. What is needed is to stimulate the development of applications in a controlled fashion to the point at which their full commercial viability gains critical mass. What is needed are relatively inexpensive controlled experiments which combine the implementation of next generation infrastructure with a focused effort to create the next generation applications prototypes. These efforts should be a very high priority not only of the NREN but of the telephone companies as well in the deployment of narrowband ISDN. One should not assume existing information providers will be the major players. In PC's existing mainframe and minicomputer software houses did not dominate PC software market. In fact, they were an insignificant factor. Existing information services providers will clearly benefit from the development of an NPN and should be included in the design and development process, but they are unlikely to develop the unanticipated new applications which will create huge new markets. There is a fertile computer underground of tens of thousands of non- commercial computer bulletin boards, electronic newsletters and other publications, chat lines, and other services which operate in a completely ad hoc fashion mostly over the public switched telephone network and to some extent over the Internet. Efforts should be made to include the designers of these grass roots experiments in digital media in the development of applications and services for the NPN. The NPN should encourage information entrepreneurship. Make it as easy to provide a service as it is to order a business telephone and get a listing in the yellow pages. The architectural design of NPN should be heavily influenced by these considerations. Now is the time to invite prospective developers in while they can influence the design of the platform. Government should consider how to accelerate commercial development by selective funding of key research prototypes of network applications. These efforts should actively attempt to include creative talent from across the entire spectrum of computing and communications technology. Recommendation #5: Design the NPN with the intent of fully applying first amendment rights of freedom of speech, freedom of the press, and freedom of assembly to its users. Among the many ultimate uses of the NPN, information and communication applications will be in the first rank. As such, our society will face many of the choices it has faced in the past with the creation of new media such as the telephony and broadcasting. As Ithiel de Sola Pool pointed out so clearly in "Technologies of Freedom", there are critical choices to be made in the early years of a new medium with regard to the regulatory model to be adopted. The lack of regulation and government ability to control print media stands in sharp contrast to the heavy regulation and control over broadcasting. The development of new digital media based on a national public network will raise these issues once again. Because digital media represent a convergence of all previous media in including elements characteristic of print, telephony and other forms of common carriage, and broadcasting, the process of developing a social consensus about the treatment of digital media is especially challenging. I would agree with de Sola Pool in recommending that the public interest will be best served by a regime which encourages the greatest diversity and hence the greatest public choice. The print model of protection of free speech through the general absence of censorship and government control, as buttressed by the first amendment, offers the greatest chance of achieving this end. Conclusion: Obviously there are an enormous number of programmatic details to be worked out to realize these recommendations. As well, many of the propositions set forth may be regarded as controversial. If this paper has succeeded in injecting new ideas into the public discourse, it must be considered successful. Mitchell Kapor, President Electronic Frontier Foundation, Inc. 155 Second St. Cambridge, MA 02141 Internet: mkapor@eff.org MCI Mail: mkapor (617) 864-1550