Economic Factors to Consider in Developing Research Goals for the HPCC Program.

Scott Bradner - Harvard University - December 1997

Developed for the Broadbased Subcommittee of the Presidential Advisory Committee on High Performance Computing and Communication, Information Technology, and the Next Generation Internet

To explore the economic factors that will shape the Internet of the future we have to make some technology assumptions. We have to assume that technology will be developed and deployed that can provide reasonable quality of service controls in very large-scale networks. It is very hard to see how a common Internet-based data infrastructure can have economic viability if this does not happen. Many of the services, which could be supported over a common data infrastructure, require a far better assurance of reliable data delivery than the current Internet can provide. Widely useful Internet-based telephony, video on demand, and electronic commerce require an infrastructure in which the characteristics of the transport of data for different applications can be controlled separately. We also have to assume the development and deployment of scaleable authentication, authorization and accounting technologies.

Assuming that the required network support technologies are developed and can be deployed, a common data infrastructure in the US may have reasonably attractive economics. This is true even though there are many organizations which would have to share in the revenue stream, but the viability could be adversely effected by a number of factors including governmental regulation.

Economic viability of a multi function Internet

There are currently a number of parallel physical information exchange systems, which could potentially share a common logical infrastructure. We say 'logical infrastructure' because a common logical data infrastructure can be carried over many physical infrastructures. The largest single infrastructure in the US is the telephone system, with about $200 billion in annual revenue (local and long distance combined) and 155 million individual phone lines. The US cable TV system with annual revenues of $3.9 billion services 59 million households. The broadcast TV networks have annual revenue of $25 billion and reach 96 million homes. The electronic data interchange (EDI) industry is in its infancy but has annual revenues of $122 million. In addition, some functions, which are currently performed manually at significant cost, could be done over a common data infrastructure, such as reading utility meters. A multi-function integrated services network should be able to support all of these functions so the potential revenue stream for a future multi-function Internet far exceeds the current $3.3 billion combined annual revenue for the 5500 US Internet service providers (ISPs). This is not to imply that there will likely be any fast migration of these services away from their current delivery methodologies.

Some functions, which were once thought to have the potential to make significant revenue contributions in the future, have not proven to be major players. For example, the pay per view market is only $660 million per year and the total video rental business is about $9 billion. The potential of others such as interactive multi-player games has yet to be proven.

It would seem that an integrated multi-function network can only be viable if it can support voice traffic with good enough quality to attract users away from the current voice network or support cable TV quality video traffic. Quality, in and of itself, is not sufficient to ensure viability. Each service on an integrated services network must be able to be offered at a lower cost point than the same service over traditional infrastructures. For example, if an integrated service network could provide a customer with good quality voice service and high-speed Internet service on the same line it would be quite attractive because many customers would not have to purchase a second phone line. In addition, packet-based multiplexing of compressed voice and compressed video can save some costs because it can make far more efficient use of network bandwidth.

Without the development of some ability for the Internet service providers to offer controlled-quality data services it is very hard to see how the Internet can generate enough revenue to support itself and still make the investments required to upgrade the infrastructure to meet customer demands unless the market is distorted by a dominate vendor. Currently ISPs can only offer best-effort Internet service and must compete with each other only on a price basis. This tends to cause the vendors to maximize the usage of their networks and thus causes the service quality to degrade to a point where extra value services are not attractive because of poor quality. The development of technology that makes it possible for ISPs to offer different levels of service guarantees will provide the basis for ISPs being able to offer more than one service, and be able to charge differently for the different types of service. Such technology will also make it feasible for the extra value services, such as Internet telephony, to be supported at reasonable levels of quality.

The cost side of the calculation is somewhat harder to predict. Competition in last-mile technologies between telephone copper loop technologies (including ISDN and various digital subscriber line technologies which are supportable over the majority of the existing telephone cable plant), cable TV coax cable, wireless, and power line transmission should keep these costs to a minimum. Even so these costs can be substantial. For example, local loop costs are currently more than 10% of the total telephone system revenue. The current cost of wide area fiber installations is also substantial, with the telephone infrastructure consisting of over 16 million miles of installed fiber. Currently this fiber is very well allocated and additional fiber is being installed rapidly and at great cost. But the difficulty of fiber installation and the effective cost of the use of this fiber should drop precipitously with the deployment of wave division multiplexing (WDM). WDM can be used to multiply the bandwidth available on an existing single fiber by factors of 16 to 64. One factor complicating the deployment of WDM by the regional telephone companies is that these companies must by regulation make excess capacity available to their competitors. Thus a regional telephone company that installs WDM technology in excess of their current requirements is, in a way, building their competitor's infrastructure--not an attractive prospect.

Impact of regulations

Currently the Internet itself in the US is largely free of regulations and taxes but this may not continue. If there is any significant migration of the existing local or long distance voice traffic to a new integrated services infrastructure there will be a great deal of pressure to migrate some of the existing telephone taxes and regulations at the same time. This is particularly true with the universal service fund. In 1997 many of the players in the telephone business paid $1.7 billion into a fund that has been used to help subsidize "life-line" phone service and to help offset the higher per-user cost of installing and operating phone service in high cost rural areas. For 1998 the FCC has expand the fund to $4.65 billion and broadened its mission to include subsidizing inside wiring and Internet connectivity for schools, libraries and rural healthcare facilities. This expansion is the first attempt to extend the concept of universal service to the Internet arena. This has been done without a clear consensus on just what universal service means in the digital age. It is easy to define universal service for simple telephones. It is the ability for someone to place and receive phone calls. The network capacity required to provide this function is well defined. But it is not so easy to define what constitutes universal service for the Internet. It is far from clear what minimum bandwidth or quality of service is going to be seen as basic Internet service. Additionally, the FCC regulations for 1989 stipulate that all of the funds for this pool will come from telephone service providers and that ISPs, even cable TV-based ones, are exempt. This could easily change in the future and have an impact on the costs and therefore the viability of providing integrated network service.

The most difficult area for regulations is migrating the regional telephone companies, which control most of the 'last-mile' infrastructure, from the regulated monopoly environment into a fair, fully competitive playing field. The exclusive control that these companies have over large parts of the telecommunications infrastructure is a major issue for the future of the Internet as well as for most aspects of the future of US the telecommunications industry.

Economic models

It has become clear over the last few years that the initial NII assumptions of a relatively few specialized content providers being the major economic drivers have not been proven out. The Internet has shown that there will be tens of thousands, if not millions of content providers, ranging in size from personal web pages to on-line newspapers and video on demand services. The future integrated data infrastructure will have to be paid for primarily by the users of services rather than by those retrieving content. The basic economic model must be one where services, such as telephony, can be delivered more efficiently over an integrated network and thus these services become attractive in the marketplace. Users should be quite willing to pay for identifiable functions like reserving network resources for a phone call if these functions are tied to specific services and the costs bundled as part of the fee for the services.

It is reasonable to expect that both subscription and per use fee structures will be offered for most network-based content providers such as on-line news services and database search engines. These services might also be able to add on a surcharge if the user requests a higher quality of service for the delivery of the content. It is also reasonable to expect that current flat rate and per use pricing models for telephony, fax and EDI services will be offered when these services are delivered over a new infrastructure.

It is reasonably safe to predict that local and backbone ISPs will use a combination of flat rate and usage-based monthly fees for access to the integrated services network and, in addition, will have flat rate and usage-based fees for traffic that requests a higher quality of service from the network. These fees will be in addition to the fees that the service providers will charge for their services. A user will have a basic monthly access fee of some sort from their network provider and additional fees from telephony and content providers.

It is quite a bit harder to predict what economic models the Bell operating companies and GTE will use. The current fees from these companies have been distorted by many years of close regulation and by the inclusion of a number of government-imposed fees and taxes. In spite of current statements to the contrary, it seems likely that if there is any significant movement of telephony services to an integrated services Internet that at least some of these fees and taxes would follow and thereby complicate the economic models of both the ISPs and the regional telephone companies.

Effects of an Internet economy

The Internet economy has the potential of remaking major components of the US and world economies. Almost all business to business non-voice communications could be operated over the Internet today at reduced costs if there were better assurances of the security of Internet traffic and predictable qualities of service over the Internet. This includes catalogues, orders, invoices, materials tracking notices, payments, technical specifications, schematic drawings, text-based communications and the many other messages that one business exchanges with another in the normal course of doing business. The migration of business to business communications can also include interactive voice and video if more advanced QoS controls are included. Cisco Systems has demonstrated the potential of the Internet economy by migrating just a part of the product ordering and tracking proces for their small to meduium business customers to the Internet. For Cisco alone this has resulted in savings of hundreds of millions of dollars per year on sales of billions of dollars per year.

Additionally, the Internet can be used for direct sales to individuals. Most of the larger PC and computer supply companies have extensive Internet-based catalogs, Amazon.com and now Barns & Noble have established on-line bookstores, each with a far larger selection of materials than could ever be housed in a traditional store. On-line sales may, over time, have an impact on the finical viability of traditional stores in some technology areas.

The IRS web site exemplifies the efficiency by which information can be offered on the Internet. Electronic delivery of tax forms can be accomplished for a fraction of a cent per form compared to more than a dollar per form in walk-in IRS offices. Electronic distribution of government information has already fundamentally changed the ability of the citizens to monitor the actions of Congress, reducing the time it takes to get a copy of pending legislation from weeks to minutes. The Security and Exchange Commission's on-line service for the retrieval of the millions of pages of corporate documents has made it significantly easier and far less expensive to do the background research necessary before investment.

The inherently borderless Internet means that potential customers from around the world can reach on-line merchants with little more effort on the part of the merchant than setting up a web page for local use. Multi-language sites can further enhance their ability to compete globally. The impact of the global nature of the Internet is magnified for vendors of software and electronic services, such as database access, stockbrokers and banking. The purchased goods can be delivered via the Internet, with no regard to tariffs or distributor territories. A backroom operation in a small country can have a significantly greater market presence than a hundred-year-old Wall Street firm.

Areas for research

A number of the major vendors of internetworking technology are working on quality of service technology but most of these efforts are aimed at getting something for immediate deployment rather than at understanding basic principals. Basic research into scalable quality of service technologies is still needed and may need to be funded by the US government. There are a number of other important areas of exploration; authentication mechanisms, authorization systems, accounting strategies and support functions, and very large-scale network management tools that can support a diversified, quality of service controlled network. None of these areas are being adequately explored in the commercial sector for networks of the scale that can be reasonably projected for the Internet in the next few years and will have to be well supported in federal research programs.