CALIFORNIA STATE UNIVERSITY AT DOMINGUEZ HILLS
The Clinton adminstration has set ambitious goals for a national and global information infrastructure (NII and GII). The administration plan for the NII was articulated by Vice President Gore in a speech in Los Angeles [5.]. He stated the administration would:
1. encourage private investment in the NII
2. provide and protect competition
3. provide open access to the network
4. act to avoid creating a society of information "haves" and "have nots"
5. encourage flexible and responsive government action.
In another speech, the Vice President called for a GII. He called upon "legislators, regulators, and business people" to build and operate a GII, which would "circle the globe with information superhighways on which all people can travel," and he pledged "vigorous, continued participation in achieving this goal." [6.]
In the U. S., both the Internet and interactive TV communities are working toward this infrastructure. While they use some of the same terms and have both received the administration's attention, these groups seem to be pursuing independent courses - - they work in different industries, attend different conferences, and have different cultures. This paper summarizes the current state of each effort, then it discusses some of the key differences between the two.
The Internet is large and growing rapidly. Between January 1993 and January 1994, the number of hosts grew from 1,313,000 to 2,217,000, a 69% increase [10.], and monthly traffic on the US NSF backbone alone is about 10 terabytes.[1.]. With an estimated 20 million users, the Internet is larger than many nations, and has a highly educated, mostly professional population.
There are many signs of increased commercialization. Most organizations which build IP internets (small i), register them with the Network Information Center, regardless of whether they connect to the Internet or not. Registrations are growing at 12% per month, and roughly 75% of those are from business enterprises. The Internet is positioned to provide the backbone which glues enterprise-wide and inter-enterprise networks together. Many businesses have been worried about the lack of security on the Internet, but we know how to install firewall computers in front of corporate networks and encryption and signature authentication are being designed and deployed. Facilities for billing are also being tested, and there are a number of experimental Internet-based "malls" and "markets." Another indicator of commercial interest is the fact that at least 4 business-related Internet newsletters began publication during the second half of 1993. Note that the U. S. government is funding development of new technology and standards, not enterprises, though there has been some controversy over the line between them.
The Internet is becoming more global. Sixty two political units are directly connected, and the international growth rate is 9% per month (in numbers of networks) [18.]. At current rates, international connectivity will exceed U. S. connectivity around October, 1994. Table 1 summarizes global connectivity and recent trends.
8-1-93 2-15-94
Full IP.........57...........................61 Email..........137..........................146 None............99...........................91 Total..........236..........................237
There are also new ways to get onto the Internet. Performance Systems International (PSI) is test-offering Internet connectivity via cable TV. Marty Schoffstall, PSI's chief technical officer, predicts a rosy future for Cable TV access to the Internet. He feels the Internet and cable company cultures are compatible -- both are "cowboys."
Wireless Internet email has been available since 1992 from RadioMail, and several vendors are introducing cellular packet- switched data services. Networking pioneer Paul Baran is developing a system for wireless connectivity over CATV. Even ISDN has been resurrected as a viable path to the office LAN from home or the road. Forrester Research interviewed 50 Fortune-500 companies, and after considering six alternatives, concluded that "(by 1995, ISDN) will be the first option to provide affordable, nationwide, near-ubiquitous, switched digital connectivity for low volume access." [2.] Around the turn of the century, we may see satellite-based access. Perhaps the most promising satellite venture is Teledesic, which plans a network of 840 low-earth orbiting ATM switches.
There is even a new Internet "off-ramp." Carl Malamud and Marshall Rose have developed software which lets Internet hosts receive email messages (text or graphic), and automatically forward them to a specified fax phone number. The message forwarder could do this as a free service (perhaps within a corporation or at a public library). Alternatively, they could obtain revenue by charging a fee or by printing a paid ad at the bottom of the fax cover sheet. There are already servers in several cities, and several companies are buying adds[1].
Finally, bulletin board systems (BBS), which began with hobbyists and community information providers, are becoming commercial, and connecting to the Internet. Boardwatch Magazine, which covers BBS, has a regular Internet section, and their annual conference has an Internet track. In a reader survey, they chose 100 top BBS, and 21 of them offer Internet connectivity or email [19.]. These are not toy systems. Several have over 10 Gbytes on line, and they average 25 dial-in lines. The largest has 35 Gbytes, and receives 5,000 calls daily on 280 lines. Many BBS specialize, with topics from community information, software, Christian values, pornography, and senior citizens to Batman.
Jack Rickard, Boardwatch Editor and Publisher, estimates that there are 53,000 publicly-accessible BBS in North America and 92,000 world wide [14.]. This includes 22,000 Fidonet BBS, which have store-and-forward communications between them. When asked about the Internet, Rickard says, "when somebody can put an ftp- server/BBS on the Internet for a couple thousand dollars, it will certainly change the network." It will also be interesting to see how being networked changes the BBS.
The Cable TV industry has accelerated the process. Cable TV began as a means of getting signals to outlying areas. Programming came to a head-end, where it was broadcast over coaxial cable in trunk lines to feeder lines to drops to homes. Amplifiers were located at quarter-mile intervals along this network. By 1992 cable ran by 97% of U. S. households and connected 61% of them [7.] (see Table 2).
1975 1980 1985 1990 1992
TV households (millions).................70.....78.......86.......93.......93 Percent of TV households served..........13.....23.......46.......59.......61 Homes passed (millions)..................23.....35.......65.......86.......91 Percent of TV households passed..........33.....45.......76.......92.......97
To increase capacity, reliability, and picture quality, the cable companies have been converting trunks from coaxial cable to optical fiber. It costs only about $50 per subscriber to run fiber to neighborhood nodes of about 1,500-2,000 homes. The next step will be the extension of fiber to nodes serving 200-500 homes. The industry estimates that the cost of converting the U. S. cable plant in this manner is about $20 billion, compared to $200-400 billion for a comparable rebuilding of the telephone plant. The major cause of this savings is that coaxial cable running to the homes would not be replaced[2].
At some point, entertainment and cable companies realized this hybrid network, coupled with improved computing and compression technology, would be capable of delivering digital movies. If they could provide movies on demand from the safety and convenience of the home, they could take over the estimated $12 billion [17.] annual video-rental business[3], justifying much of the needed investment. Throw in a share of the $70 billion catalog shopping and $4 billion video game markets [17.] and the $2 billion already sold through TV shopping [16.], and the "interactive TV" community was born.
Silicon Graphics (SGI) is designing the video-on-demand hardware and software, and Jim Barton, Vice President of their Media Systems Division, described the project. The head-end hardware will be a cluster of 14 SGI servers on an FDDI ring. Two of the machines in the cluster will function primarily as session managers and transaction processors, and the others will deliver MPEG-compressed video. The video servers have 1.2 Gbit/s backplanes, and can have up to 24 processors, 12 Gbyte memory, and several hundred Gbytes of disk. The FDDI ring will be used for control and coordination within the cluster. Perhaps 50 popular movies will be on-line at all times, with, the top 10 or so on multiple servers. Another 5-10,000 movies will be available on robot-accessed tape, with a few minutes lead time for mounting. Small video segments will be transferred from the video servers to an AT&T ATM switch, and sent over 1.5 mb/s, full-duplex lines to homes.
The network will terminate in TV set-top boxes designed by SGI. The design is foreshadowed in a SIGGRAPH paper by Jim Clark, the Chairman of SGI [3.]. Clark speculates on a $200 (mid 1990s) box with MPEG decompression, encryption, display-device independence, and built-in graphics and image processing capability. The initial prototype will be a modified SGI Indy workstation with a 100 mhz MIPS R4000 processor and a Scientific- Atlanta add-in board for decompression and analog signal processing. There will be no storage, and relatively little memory. Bootup and client software management will be handled at the head-end, reducing system cost. For those who wish to go beyond the basic box, there will be PCMCIA slots.
There are other interesting video prototypes under way. For example, Oracle is assuming a less intelligent set-top box, and designing a video server using an nCube multiprocessor system. According to John Kish, Oracle's Senior Vice President, Business Development, they will test it in Omaha with U. S. West as a partner. Kish says Oracle plans to port their video-server software to many machines, including SGI's, but they have decided on a 512-processor nCube machine for this test. (Oracle president Larry Ellison is a major investor in nCube, but Kish insists the selection was made independently). Kish expects the 512-processor machine to serve 3,000 video streams, and says the technology scales well. He envisions serving 100,000 video streams in a few years, and says the cost of capital equipment must be around $200 per stream for movies to pay off. Oracle and SGI have very different architectures -- isn't competition nice?
If that is what home shopping becomes, I won't spend much time browsing in the video mall. But, if the NII were to increase the efficiency of the market for a variety of goods and services, it would have significant economic impact (as it already has in capital markets).
Theoretically, an efficient market provides information about goods and services, and rational, well-informed consumers make optimal decisions. However, Herbert Simon received a Nobel Prize in economics for demonstrating that such perfection does not exist in the real world. Information is not perfect, and it costs. Furthermore, shoppers have finite cognitive skills and time, so they do not optimize, they merely "satisfice," making decisions which satisfy their constraints. The NII could increase market efficiency by providing both better information and analysis tools.
Let us consider relatively large purchases first, for example an auto, appliance, or some office or industrial equipment. (Items you might screen on-line, and eventually purchase in person). One could imagine a comparative shopping service (more like a database than virtual travel) with information on product specifications, prices, delivery times, and guarantee terms. There could be video presentations by manufacturers, independent product reviews, comments by satisfied and dissatisfied customers (with their email address), and discussion groups among consumers and producers. There could also be information to allow comparison of local distributors or service providers. Such information would ease Simon's first market limitation. The second constraint, the consumer's finite rationality, could be relaxed with tools for discovery, analysis and presentation -- agents, models, and expert systems.
The NII also has the potential to increase market efficiency for staple goods. I am more curious about the effect of the NII on supermarkets than on boutiques in malls. It would be convenient to ask "what will it cost me to purchase the following grocery list from various stores near me this week." Or to see current prices for a specific computer or program. This sort of purchase will be concluded on line, and the line between retailing and distribution may shift. I really want to know what it will cost to get the items on the grocery list to my home, not which store or stores to buy it from. Of course information products like news, software, and music can be both browsed and distributed on line.
While the interactive TV community has the publicity lead on shopping, shopping might be a better fit with the Internet. Interactive TV people envision mass markets with relatively few vendors. The temperament of the Internet community points more toward a bazaar than a mass market, with easy entry and many vendors. The Internet community also has experience with the information retrieval and analysis tools which will underlie the electronic market.
All of this is a bit glib. The interactive TV community is starting trials, but widespread rollout will take many years. Technical innovation and standards are needed, and these will be followed by a long rollout period, which will begin in affluent areas. The non-technical questions are more difficult. For example, will there be alternative comparison shopping services for a class of item? Will comparison shopping services charge consumers, be funded by vendors, or be provided as a common good. The latter possibilities raise questions of regulation of goods markets, analogous to present regulation of security markets. What steps would be taken to insure accuracy? What would be the legal liabilities and procedures? Who would be able to list products?
Much of the initial vision and funding for the Internet was due to J. C. R. Licklider, the first head of the Information Processing Techniques Office at ARPA (the Advanced Projects Research Agency of the Department of Defense). Licklider sketched an outline for what he called an "intergalactic" network, and was influential in providing funding for the ARPAnet, which evolved into the Internet. The application Licklider envisioned was support for a community of scholars:
What will on-line interactive communities be like? In most fields they will consist of geographically separated members, sometimes grouped in small clusters and sometimes working individually. They will be communities not of common location, but of common interest. [9.](30)
The first applications were remote login and file transfer. Electronic mail was soon added [11.]. These are still the basic Internet services, but an extra level of tools to integrate distributed servers, e. g., Gopher, Wide Area Information Servers (WAIS), and World Wide Web (WWW), have increased efficiency and ease of use.
Internet and interactive TV applications are different, because they serve different users. Interactive TV is targeting the consumer at home, and (in the U. S.) the Internet has broadened in focus from scholars and computer scientists to include students at all levels and business people. This diversification seems likely to continue in light of the U.S. government having called for connectivity for all classrooms, libraries and clinics. [5.]
The Internet began with Teletypes, but has really been built around desktop computers. As connectivity bandwidth increases, and the cost of digital electronics declines, it will be possible to use one terminal device on the desktop or from the couch. The hardware differences between TV sets and computers may eventually disappear, but the attitude and viewing positions are quite different [12.].
In spite of the impressive global spread of the Internet, there are some sobering concerns [4.](13). About one fifth of the world's population has access to far more capability than the rest. There is a direct correlation between measures of national development and the quality of network services available to advanced industrial and less developed countries (LDCs) [15.]. The connectivity to an LDC may be as little as a Fidonet link to a few PCs with a handful of regular users. Even within the most advanced, well connected countries the majority of the populations have little or no participation. For example, despite the much proclaimed connectivity of U. S. universities, where access is almost a free good, Internet utilization is generally confined to a minority of faculty and students. Going beyond universities, U. S. connectivity is concentrated in urban areas and is relatively sparse in regions such as the Midwest and in poor regions and neighborhoods. These discrepancies are most probably greater in developing nations.
The interactive TV community originally thought of using an augmented telephone network, but is now also heavily involved with cable TV. The differences between telephone and cable TV companies may, in the long run, become blurred as they begin offering competing services, but for now, they have different experience.
The telephone companies have a tradition of providing a highly- reliable, automatically transaction-billed service, and seamless global interoperability. By contrast cable companies typically do flat monthly billing, tolerate service interruption, and serve limited areas. Cable companies have a high level of penetration to U. S. homes, but do not typically serve business. The phone companies cover business and homes. (About 80% of U. S. homes fall within 18,000 feet of a switching office, and even more businesses do). The telephone companies are also used to universal access and symmetrical, two-way transmission. Finally, the telephone companies are (for now) profitable and wealthy. AT&T is diversified, selling ATM switches and video servers, buying related companies, and talking about linking local cable providers. The phone companies are also testing "asymmetrical digital subscriber lines," a technique for delivering 1.5 mbit/sec compressed video (VCR quality) over copper wire, though many feel this is at best an interim solution.
Open access is one of the U. S. administration's goals for the NII, and low-cost access, with every user a potential publisher, is a basic tenet of the Internet. The motivating applications were file transfer and remote login, and any host on the original ARPANET could provide either service. These are reflected in the FTP and Telnet services offered by IP-connected Internet hosts today. Even the design and implementation of the Internet protocols is done using the open, democratic mechanism of published Requests For Comment, which may originate with anyone in the community.
While Internet information providers began with FTP archives, the rapid acceptance of WAIS, Gopher, and World Wide Web attest to the ongoing assumption that everyone on the Internet is a potential publisher. I would imagine that nearly every U. S. university publishes gopher or FTP archives and sponsors discussion lists. (I personally maintain two archives and have initiated many listservers). Many of these publications are of relatively local interest, but the Internet-wide "New List" and "Net Happenings" lists include well over 100 announcements per week.
We have been speaking of access by publishers, but access by users is also an issue. As we have seen, the U. S. administration hopes to avoid a gap between those who have and do not have network access[5]. Regulatory and funding pressure will probably make this an issue for both the Internet and interactive TV communities.
The Internet began in the education and research community. When the National Science Foundation (NSF) began providing backbone connectivity in the U. S., acceptable use policies prohibited commercial activity; however, it was planned that NSF would eventually withdraw, and there would be a transition to commercial use. As we saw above, this is occurring. Still, the growing appearance of commercial activity on the Internet is relatively new, and there remains a large constituency focusing on education and community service. The Internet community debates the issue of usage-based charges; the interactive TV community assumes them.
This difference in economic orientation also grows from the attitudes of the two communities toward the price of information. The Interactive TV culture is based on the sale of copywritten information for royalties, while the Internet grew from the academic community which values information sharing. (Though both reward early publication).
If we extrapolate current trends, we can expect that by that time, the Internet will have grown significantly, will have a higher percentage of commercial utilization, and will support a wider variety of data types and users. On the other hand, it may still be poorly suited for the distribution of movies and other connection-oriented communication.
I must admit to a degree of discomfort with the interactive TV community. After all, it is headed by the folks who bring us "Cops," reruns of "Gilligan's Island," and commercials for 1-900 numbers and "workman's compensation" lawyers. Do we want them in control of the deployment and application of a technology that may have as much impact on society and our worldview as the clock or electric light?
Regardless of who is in charge and what their motivation is, NII technology will have profound effects -- good and bad, foreseen and unanticipated. I already see this in my own work, where daily collaboration with remote colleagues would be impossible without the Internet. On the other hand, my time is finite, so I have significantly reduced professional collaboration and social interaction with people at my physical workplace. My productivity has been enhanced markedly, but my daily working relationships and organizational loyalty have shifted. The shift from local to remote is more pronounced when we consider the mass media. We know Johnny Carson better than we know our neighbors, and we are touched more deeply by televised events than those on our own blocks.
Working and playing in cyberspace will change our relationship to time, abstraction, and people. It will change our values, perception, and addictions. Regardless of who builds the NII, let's keep our eyes open.
[2.] Batson, J. and Hyland, J. L., "Don't Laugh, It's ISDN," The Network Strategy Report, vol. 7, no 7, pp 2-13, Forrester Research, Cambridge, MA, June, 1993.[Retour.]
[3.] Clark, J., "A Tele-Computer," Proceedings of SIGGRAPH '92, pp 19-23, Addison-Wesley, Reading, MA, 1992. [Retour.]
[4.] Goodman, S, Press, L, Ruth, S., and Rutkowski, A., "The Global Diffusion of the Internet: Patterns and Problems," Communications of the ACM, in press.[Retour.]
[5.] Gore, Al, "Remarks at Royce Hall, UCLA," Los Angeles, January 11, 1994. Anonymous ftp, sunsite.unc.edu, pub/academic/political-science/internet-related/VP-Speech-at- UCLA[Retour.]
[6.] Gore, Al, "Remarks Prepared for Delivery, International Telecommunications Union," Buenos Aires, March 21, 1994. Anonymous ftp, sunsite.unc.edu, pub/academic/political- science/internet-related/Gore's-Delivered-Remarks-on-NII-12- 21-93[Retour.]
[7.] Green, R. R., Testimony before the Subcommittee on Technology, Environment, and Aviation, Committee on Science, Space, and Technology, U. S., House of Representatives, March 23, 1993.[Retour.]
[8.] Landweber, Larry, "The Connectivity Table," Version 10, February 15, 1994, anonymous ftp, ftp.cs.wisc.edu, connectivity_table directory.[Retour.]
[9.] Licklider, J. C. R. and Taylor, Robert W, "The Computer as a Communication Device," Science and Technology, April, 1968, 21-31.[Retour.]
[10.] Lottor, Mark, Report-9401, directory pub/zone, ftp.nisc.sri.com, January, 1994.[Retour.]
[11.] Marill, Thomas and Roberts, Lawrence G., "Toward a Cooperative Network of Time-Shared Computers," Proceedings of the 1966 Fall Joint Computer Conference, 425-431. [Retour.]
[12.] Press, L., "Compuvision or Teleputer?" Communications of the Association for Computing Machinery, September, 1993, pp 66- 80.[Retour.]
[13.] Press, L., "The net: progress and opportunity," Communications of the ACM, 35, 12 (December 1992), 21-25.
[14.] Rickard, Jack, and Funk, Gary, "The International FidoNet -- 22067 Bulletin Boards with an Attitude," Boardwatch, August, 1993, pp 66-80.[Retour.]
[15.] Ruth, S., Goet, R. "Must invisible colleges be invisible? An approach to examining large communities of network users," Internet Research, 3, 1 (Spring 1993), 47. [Retour.]
[16.] Zinn, L, De George, G., Shoretz, R., Yang, D. J., and Forest, S. A., "Retailing will never Be the Same," Business Week, July 26, 1993, p. 54-60.[Retour.]
[17.] Zoglin, Richard, "The Info Highway," Time, April 12, 1993, pp 56-61.[Retour.]
[18.] -- "Internet global statistics: 1993," Internet Society News, 2, 4 (Winter 1994), 7.[Retour.]
[19.] --, "Boardwatch-100, Readers' Choice BBS Contest Results," Boardwatch Magazine, September, 993, pp 68-80. [Retour.]
[20.] --, "Status Report on Two-Way Testbeds," Cable Television Information Center, Washington, D. C., May, 1973. [Retour.]
[2] The statistics and estimates in this section were taken from [7].[Retour.]
[3] Twenty percent of this revenue is from late charges on overdue rentals, which would not occur with video on demand.[Retour.]
[4] Two-way analog tests date back to the early 1970s. One of the earliest was also in Orlando, conducted by Orange Cablevision [20.].[Retour.]
[5] This possibility was foreseen from the start. Licklider and Taylor wrote:
Will 'to be on line' be a privilege or a right? If only a favored segment of
the population gets a chance to enjoy the advantage of 'intelligence
amplification,' the network may exaggerate the discontinuity in the spectrum of
intellectual opportunity. [9.](31).[Retour.]