Berkeley Technology Law Journal
`
`Volume 14 | Issue 2
`
`March 1999
`
`The Speed Gap: Broadband Infrastructure and
`Electronic Commerce
`Howard A. Shelanski
`
`Article 10
`
`Follow this and additional works at: http://scholarship.law.berkeley.edu/btlj
`
`Recommended Citation
`Howard A. Shelanski, The Speed Gap: Broadband Infrastructure and Electronic Commerce, 14 Berkeley Tech. L.J. 721 (1999).
`Available at: http://scholarship.law.berkeley.edu/btlj/vol14/iss2/10
`
`Link to publisher version (DOI)
`http://dx.doi.org/doi:10.15779/Z38R953
`
`This Article is brought to you for free and open access by the Law Journals and Related Materials at Berkeley Law Scholarship Repository. It has been
`accepted for inclusion in Berkeley Technology Law Journal by an authorized administrator of Berkeley Law Scholarship Repository. For more
`information, please contact jcera@law.berkeley.edu.
`
`Netflix 1014 - Page 1
`
`

`
`ELECTRONIC COMMERCE SYMPOSIUM
`
`THE SPEED GAP: BROADBAND INFRASTRUCTURE
`AND ELECTRONIC COMMERCE
`
`By Howard A. Shelanskit
`
`ABSTRACT
`
`Although high-speed, broadband telecommunications services are
`not yet widespread outside of urban and commercial areas, they are
`starting to reach an increasing range of residential customers. Greater
`availability of high-speed communications links is likely to increase the
`growth of electronic commerce and other Internet applications, to the
`benefit of consumers and online businesses alike. Regulation of ad-
`vanced services may, however, affect the speed of residential broadband
`deployment and the prices for such services in the short run. This essay
`discusses some important legal constraints underlying current regulatory
`proceedings and the impact those constraints may have on the spread of
`affordable broadband services.
`
`TABLE OF CONTENTS
`
`1.
`
`AN OVERVIEW OF BROADBAND AVAILABILITY TO CONSUMERS ............................ 722
`A. Current Deployment of Advanced Network Capability ................................. 723
`B. Broadband Options in the "Last M ile". .........................................................
`724
`1. Telephone Network Solutions: ISDN and DSL .................................. 725
`727
`2. Cable Network Solution: Cable Modems ...........................
`3. W ireless and Satellite Solutions .............................................................
`728
`4. Where the Residential Market Is-and Where It Needs To Be ............... 729
`THE IMPORTANCE OF BROADBAND CONNECTIONS FOR E-COMMERCE ................... 731
`A. The Consumers' Perspective: Lowering Search Costs .................................. 731
`B. The Sellers' Perspective: Reducing Barriers to Entry ....................................
`732
`C. The Advantages of Broadband and the Challenge for Telecommunications.735
`III. REGULATION OF BROADBAND SERVICES AND IMPLICATIONS FOR E-COMMERCE...736
`A. Background of the Advanced Services Proceedings ...................................... 736
`
`11.
`
`© 1999 Howard A. Shelanski.
`f Acting Professor, University of California, Berkeley, Boalt Hall School of Law
`(on leave 1998-1999); Senior Economist, Council of Economic Advisors. I am grateful
`to Bert Huang, the editors at the Berkeley Technology Law Journal, and to participants in
`the conference on the Legal and Policy Framework for Global Electronic Commerce,
`March 5-6, 1999 at the University of California at Berkeley. The views expressed in this
`essay are the author's and are not necessarily shared by the Council of Economic Advis-
`ers or any other government agency.
`
`Netflix 1014 - Page 2
`
`

`
`BERKELEY TECHNOLOGY LAW JOURNAL
`
`[Vol. 14:721
`
`B. The FCC's Advanced Services Proceedings ..................................................
`C. The Regulatory Outlook: Increased Competition at the Cost of Cheaper
`7 39
`S p eed ? ............................................................................................................
`IV . C ONCLU SIO N ........................................................................................................... 744
`
`738
`
`Telecommunications infrastructure is critical to the growth of elec-
`tronic commerce. Telephone networks, cable systems, and other providers
`of facilities are essential intermediaries that can shape the volume and na-
`ture of transactions between online buyers and sellers. The faster and less
`expensive the links are between users and the Internet, the more quickly
`electronic commerce is likely to grow. Competition, innovation and regu-
`latory changes have all contributed to the development of a more efficient,
`higher capacity telecommunications network that is increasingly well
`suited to moving large amounts of data quickly. There is, however, a point
`at which broadband transmission stops: the local, residential network. The
`extension of broadband capability beyond its current scope to a majority
`of small businesses and households is an important challenge for the
`communications industry.
`Part I of this essay will discuss the current state of broadband capabil-
`ity in U.S. telecommunications networks. Part II will then discuss the im-
`portance for electronic commerce of increasing residential access to ad-
`vanced, high-speed telecommunications services. Finally, Part III will ex-
`amine how statutory constraints and tradeoffs underlying current regula-
`tory proposals might affect the availability and affordability of residential
`broadband services. It suggests that the 1996 Act may constrain the Fed-
`eral Communications Commission ("FCC") to favor rules that maximize
`the number of competitors in the broadband market at the expense of rules
`that maximize the spread of low-priced, advanced service offerings to
`residential customers.
`
`I. AN OVERVIEW OF BROADBAND AVAILABILITY TO
`CONSUMERS
`This section will begin by discussing changes in the telecommunica-
`tions system's ability to provide high-capacity lines to customers and to
`process information in digital format, both of which are essential for
`broadband services. It will then discuss how, because of the high costs of
`deploying fiber lines to most individual customers, several technologies
`have been developed to increase capacity of the communications plant that
`telephone and cable carriers have already constructed. It will argue that
`deployment of those technologies-namely integrated services digital
`network ("ISDN"), digital subscriber line ("DSL"), and cable modem
`
`Netflix 1014 - Page 3
`
`

`
`1999]
`
`THE SPEED GAP
`
`service-has helped to make broadband service cheaper and more widely
`available, but not yet on a ubiquitous scale to residential consumers.
`A. Current Deployment of Advanced Network Capability
`Substantial progress has been made in upgrading telecommunications
`infrastructure to meet the needs of the information sector of the economy.
`When AT&T was broken up in 1984, not one "central office"-the offices
`where the switches that route telephone calls are located-contained ad-
`vanced, digital signaling technology. By 1997, over 97 percent of central
`offices deployed such technology,1 and over 99 percent of customer lines
`were routed through such switches.2 Similarly, in 1984 only a very small
`number of links used to transport telephone traffic between central offices
`were made of fiber optic cable; the vast bulk were low-capacity copper
`lines.3 By 1990, 60 percent of interoffice transmission links were fiber,
`and by 1997 the proportion of fiber transport plant had reached nearly 96
`percent.4
`FCC figures show that from 1993 through 1997, overall deployment of
`high-capacity, fiber optic cable in the U.S. telephone system increased
`from 2.3 million to 3.4 million miles in long-distance networks,5 from 6.6
`million to 12.2 million miles in incumbent local telephone networks, 6 and
`from 0.2 million to 1.8 million miles in competitive local exchange net-
`works. 7 Total fiber mileage increased an estimated 16 percent in 1997
`alone, and actual fiber capacity by the end of 1998 was almost certainly
`much higher.
`While the paving of the "Infobahn" has reached the freeways and main
`roads, it has not yet reached the neighborhood streets. For the most part,
`the high-capacity fiber infrastructure stops well short of individual cus-
`tomer lines-often called "loops," or the "last mile"-that connect indi-
`vidual customers to the network. Of the 150 million customer lines oper-
`
`1. See INDUSTRY ANALYSIS Div., FEDERAL COMMUNICATIONS COMM'N, TRENDS
`IN TELEPHONE SERVICE 90 tbl.17.2 (July 1998).
`2. See id.
`3. Indeed, in 1986, total fiber deployment by AT&T was less than 30 percent of its
`total network, including long distance lines where the bulk of fiber was used. See John
`Haring & Ewan Kwerel, Competition Policy in the Post-Equal Access Market, 62 Rad.
`Reg. 2d (P & F) 587, n.18 (OPP Working Paper, Feb. 1987).
`4. See INDUsTRY ANALYSIS Div., supra note 1, at 91 tbl.17.3.
`5. See JONATHAN M. KRAUSHAAR, FEDERAL COMMUNICATIONS COMM'N, FIBER
`DEPLOYMENT UPDATE END OF YEAR 1997 10 tbl.2 (1998).
`6. See id. at 24 tbl.6.
`7. See id. at 36 tbl.14.
`
`Netflix 1014 - Page 4
`
`

`
`BERKELEY TECHNOLOGY LAW JOURNAL
`
`[Vol. 14:721
`
`ated by the Bell operating companies (the major incumbent carriers), 86
`percent were copper and only 14 percent were fiber at the end of 1997.8
`Because some competitive local exchange carriers have been building all-
`fiber networks, the percentage of fiber loops for the overall market may be
`slightly higher than the percentage for the incumbents' networks alone.
`But the competitive carriers have only about 3 percent of the local market
`by lines,9 so the total percentage of customer lines served by fiber loops is
`still almost certainly under 20 percent.
`Not only is the proportion of fiber loops small, but the distribution of
`those links is heavily skewed toward businesses and urban customers.
`Once fiber "backbones" are put in place in dense areas, as they have been
`in many cities, it can be economical to build a fiber link from the back-
`bone to an office or apartment building. The distances are short-often a
`matter of yards-and a single building will either have multiple customers
`or a very high-revenue customer. The economics of building fiber links to
`customers in less dense areas are much less promising. Loops are much
`longer-a matter of miles rather than yards-and at the end of that loop
`generally lies one, relatively low-revenue customer. As a result, no carri-
`ers are currently building fiber lines to individual customers outside of the
`densest urban areas. °
`B. Broadband Options in the "Last Mile"
`The absence of fiber deployment to individual customers means that
`the speed of data transport drops precipitously at the point where informa-
`tion is handed off from the network's transport lines to the customer's
`loop. Given the time and cost required to build out fiber networks, the so-
`lution for bringing broadband service to residential customers must, in the
`foreseeable future at least, work over existing residential infrastructure:
`either the copper phone loops or the coaxial links of the cable television
`network. In addition to solutions based on landline telephone and cable
`systems, wireless technologies may also become important in the residen-
`tial broadband market. Today, three technologies that meet the constraints
`of existing facilities are beginning to enter the market for residential
`
`INDUSTRY ANALYSIS Div., supra note 2, at 91 tbl.17.3.
`8.
`9. See COUNCIL OF ECONOMIC ADVISORS, PROGRESS REPORT: GROWTH AND
`COMPETITION IN U.S. TELECOMMUNICATIONS 1993-1998, 24 (Feb. 8, 1999) (White Pa-
`per).
`10. See KRAUSHAAR, supra note 5, at 21 n.18.
`
`Netflix 1014 - Page 5
`
`

`
`19991
`
`THE SPEED GAP
`
`broadband: ISDN line service and DSL service over the telephone net-
`work,' 1 and cable modem service.
`1. Telephone Network Solutions: ISDN and DSL
`Two ways of providing broadband transmission over copper telephone
`lines at modest cost are now in use. These technologies, ISDN and DSL,
`differ in their capabilities and in how they make use of existing infra-
`structure. ISDN allows transmission rates up to 128 kilobytes per second
`(kbps) over the circuit-switched voice network, about twice the best rate
`achievable by conventional modems. 12 Using an ISDN modem is just like
`using a regular computer modem in that each use requires dial-up to the
`telephone network. According to FCC data, by 1997 about 40 percent of
`local telephone company central offices, where the main switches that
`serve customer lines are located, were capable of providing ISDN serv-
`ice. 13 Those central offices together serve about 93 million customer lines,
`roughly 70 percent of the total in the United States. 14 Residential ISDN
`prices have recently fallen to as low as $25 per month (not including
`Internet access), with initial set-up charges of $125 plus the cost of an
`ISDN modem.'5 ISDN's drawbacks include potentially high usage pay-
`ments, frequent difficulty in achieving maximum bit rates, and the lack of
`an "always-on" connection that can be used without the delay of a dial-up
`process. 16
`DSL service overcomes some of the drawbacks of ISDN because it
`bypasses the circuit-switched voice network by routing data traffic to a
`packet-switched network. This allows more economical always-on con-
`nections and much faster speeds. By using modems that divide copper
`phone lines into separate bands for data traffic, DSL achieves download
`
`11. At the high end of the telecommunications market are high-capacity data links
`called T1 (or T3) lines. Prices vary by distance, contract length, and share of line capac-
`ity, with the minimum monthly charge being around $300. See Telco Express (visited
`Mar. 2, 1999) <http://digiquote.telcoexpress.com> (providing an online pricing tool for
`digital line rates around the country based on location and distance). Because TI lines
`tend to be affordable only for large businesses and institutions, they are not considered
`part of the solution for real consumer-level broadband service-i.e., service affordable by
`households and small businesses.
`12. See Digital Starter: ISDN, COMPUTER SHOPPER, Feb. 1999, at 300.
`13. See INDUSTRY ANALYSIS Div., supra note 1, at 90 tbl.17.2.
`14. See id.
`15. See Digital Starter, supra note 12, at 300. Usage charges are I to 2 cents per
`minute in addition to the monthly fee.
`16. See Richard Sekar, A Panacea for DSL Access, TELEPHONY, Jan. 18, 1999.
`
`Netflix 1014 - Page 6
`
`

`
`BERKELEY TECHNOLOGY LAW JOURNAL
`
`[Vol. 14:721
`
`speeds from 128 kbps to 7 Mbps. 17 DSL service is not yet widely avail-
`able, but that is changing. By the middle of 1998, DSL service was avail-
`able to at least some consumers in about 30 states,' 8 and various providers
`have announced aggressive plans to expand the reach of their DSL offer-
`ings. For example, incumbent local exchange companies are pursuing dif-
`ferent strategies, but are aiming to serve between 24 and 70 percent of
`their customers by 2000.19 In addition, competitive local exchange com-
`panies focusing on data services have entered a number of markets. Alto-
`gether, independent analysts predict that by 2000, over 40 million U.S.
`households will have access to DSL service.20
`Prices for DSL have started to fall accordingly. Bell Atlantic offers
`DSL service with Internet access at prices starting as low as $40 per
`month, plus an installation charge recently listed at over $400. Pacific
`Bell now offers DSL service, including Internet access, for as low as $39
`per month for 384 kbps speeds; installation and necessary equipment re-
`quire an additional one-time fee of just under $200.22
`Although DSL is promising and becoming more widely available, sev-
`eral technical issues limit the number of customers with access to the
`
`1999)
`
`17. See Shawn P. McCarthy, Internet Technologies to Watch, LOGISTICS MGMT.
`DISTRIBUTION REP., Jan. 31, 1999, at 74.
`18. See Memorandum from Carol W. Wilner, Director, Federal Government Rela-
`tions, AT&T, to author, (Feb 5, 1999) (on file with Berkeley Technology Law Journal)
`(providing maps depicting states with DSL and cable modem service).
`19. SBC, which currently serves 37 million customers, is targeting 8 million for
`broadband availability by 2000. See Got Bandwidth? Pacific Bell Answers California's
`"need for speed" with $39 ADSL Service, Major Availability (visited Apr. 9, 1999)
`<http://www.sbc.com/PB/News/Article.html?query-type=article&query= 19990112-04>.
`Bell Atlantic, which serves 42 million customers, has set approximately the same goal.
`See Bell Atlantic.net cuts price of Infospeed DSL package (visited Apr. 9, 1999)
`<http://www.ba.com/nr/1999/Mar/19990331001.html>. Bell South began service in 1998
`with 7 cities, and plans to offer service in 30 cities total by 2000. See Fastaccess city
`(visited Apr.
`9,
`1999)
`<http://www.bellsouth.net/extemal/adsl/
`availability
`city-availability.html>. U S West currently provides service in 40 cities. See U S WEST
`Brings Lightning Fast New Internet Access to Homes in 40 Cities by June 1998 (visited
`Apr. 9, 1999) <http://www.uswest.com/com/insideusw/news/012998.html>. Ameritech
`plans to make broadband available to 70 percent of its 21 million subscribers by 2000.
`See Ameritech Launches High Speed Internet Service (visited Apr. 9, 1999)
`<http://www.ameritech.com/media/release/view/0,1038,842 l2,00.html>.
`20. See Wilner, supra note 18.
`21. See Bell Atlantic
`Infospeed DSL Pricing (visited Mar. 2,
`<http://www.bell-atl.com/adsl/more-info/pricing.html>.
`22. See Fastrak DSL--Pricing & Availability
`(visited Mar.
`<http://www.pacbell.com/products/business/fastrak/dsl/pricing.html>.
`
`2,
`
`1999)
`
`Netflix 1014 - Page 7
`
`

`
`1999]
`
`THE SPEED GAP
`
`service: transmission over DSL lines is generally effective only for cus-
`tomers located a short distance, generally within about three miles, from a
`central switching office.23 Performance of DSL transmission declines with
`loop length, but also varies with condition of the loop and quality of
`equipment attached to the loop; older copper loops that have been patched
`and repaired over the decades will often have to be reconditioned before
`they are suitable for DSL transmission.24 Technological advances are
`starting to provide improvements, but for now DSL remains an option
`primarily in areas where loops are short and in good condition.
`2. Cable Network Solution: Cable Modems
`The hybrid fiber-coaxial plant of cable systems also has broadband ca-
`pacity and can be configured for two-way, high-speed data service through
`25
`the use of cable modems. As originally built, however, that pipe runs
`one-way, toward the consumer; to provide broadband service over the ca-
`ble network, the plant must be upgraded to two-way capability for the
`more interactive applications of the Internet or for voice services.2 6 The
`investment for such upgrades is substantial, and by one estimate only
`about 15 percent of systems have been converted. 27 But the natural high-
`speed capacity of cable systems, and the fact that cable is readily available
`to 98 percent of American households, make it a natural and, for resi-
`dences, the leading broadband competitor.
`Cable systems currently provide high-speed data services to about
`300,000 customers, but are expanding aggressively. By the middle of
`1998, cable modem service was available to some households in 44
`2
`28
`states. Since 1995, $18 billion have been invested in cable upgrades,29
`and it is predicted that cable modem service will be available to over 40
`2000.30 AT&T's merger with Tele-
`households
`by
`million
`Communications, Inc. ("TCI") is premised on upgrading TCI's cable sys-
`
`23. See, e.g., Eric Krapf, Slow roll for DSL, BUS. COMM. REV., Aug. 1998, at 47.
`24. See id.
`25. See generally GEORGE ABE, CISCO SYSTEMS, RESIDENTIAL BROADBAND 180-90
`(1997) (discussing the principles of operation for cable modems).
`26. To be sure, the "upstream" channel away from the consumer need not be as big
`as the downstream channel, but some upstream capacity is necessary.
`27. See James B. Speta, Handicapping the Race for the Last Mile 27 (Feb. 16, 1999)
`(unpublished manuscript, on file with Berkeley Technology Law Journal).
`28. See Wilner, supra note 18.
`29. See id. (citing Credit Suisse First Boston, The Infrastructure Report, Dec. 15,
`1998).
`30. See id. (citing various analyst reports from the Yankee Group).
`
`Netflix 1014 - Page 8
`
`

`
`BERKELEY TECHNOLOGY LAW JOURNAL
`
`[Vol. 14:721
`
`tems to serve up to 18 million customers with high-speed Internet access
`within the next few years. 31 Recently, prices for cable modem service have
`fallen to about $40 per month excluding Internet access. 32
`3. Wireless and Satellite Solutions
`Finally, wireless solutions may also be just over the horizon. The
`wireless services that are likely to provide broadband data capability are
`not, however, the cellular telephone and personal communications service
`("PCS") technologies with which most consumers are familiar. Even with
`digital conversion of the wireless telephone networks in the U.S. over the
`past several years, the data rates those systems support are less than the
`copper, landline network. 33
`More promising for broadband purposes are land-based (as opposed to
`satellite), fixed wireless systems like multichannel multipoint distribution
`service ("MMDS") and local multipoint distribution service ("LMDS").
`These systems use microwave transmission technology to send signals
`over a 30-70 kilometer radius. 34 They have the advantage of low start-up
`costs35 and by 1997 there were 73 MMDS operators serving 1 million
`video customers in the United States.3 6 MMDS and LMDS systems have
`they require line-of-sight transmission paths and are
`some drawbacks:
`subject to interference-even from bad weather. 37 MMDS is the more es-
`tablished of the two systems, and is estimated to pass over 30 million
`homes, although only about 1 million subscribe to MMDS video serv-
`
`9,
`
`1999)
`
`(visited Apr.
`
`complete merger
`31. See AT&T and TCI
`<http://www.att.com/press/item/0, 1193,382,00.html>.
`32. See Scott Bernard Nelson, Life on the Internet Fast Lane, KIPLINGER'S PERS.
`FIN. MAG., Jan. 1999, at 117.
`33. See, e.g., Douglas N. Knisley, et al., cdma2000: A Third Generation Radio
`Transmission Technology, BELL LABS TECH. J., July/Sept. 1998, at 65 (noting that the
`current generation CDMA technology-a radio transmission standard for PCS service-
`can provide data speed transmission of under 14.4 kb/s at best).
`34. See ABE, supra note 25, at 343.
`35. See id. at 347 (noting that in Los Angeles, a single MMDS antenna can reach
`upward of 4 million households, making the infrastructure investment less than $20 per
`residence in the coverage area).
`36. See id. at 344.
`37. See, e.g., In re Annual Assessment of the Status of Competition in Markets for
`the Delivery of Video Programming, FCC CS Docket 98-102, para. 81 (Dec. 23, 1998)
`(discussing line-of-sight problems) [hereinafter Annual Assessment of the Status of
`Competition]; ABE, supra note 25, at 346 (noting MMDS is limited by line-of-sight con-
`siderations); Speta, supra note 27, at 31 (citing Scott Seidel, Broadband Wireless Serv-
`ices: In the Line of Sight, Bellcore Exchange, Spring 1997, at 21-22) (noting that rain can
`affect LMDS service quality at certain spectrums).
`
`Netflix 1014 - Page 9
`
`

`
`1999]
`
`THE SPEED GAP
`
`ices. 38 Its ubiquity is promising, however, and could make MMDS an im-
`portant broadband entrant if digital compression allows its capacity to in-
`crease and if interference and other technical issues can be resolved.
`LMDS is of much more recent vintage and, although capable of very high-
`bandwidth transmission, is not considered a near-term entrant into the
`residential broadband market. 39
`Finally, satellite services have entered the market and may, as they
`have in the video market, 40 prove a powerful competitor for broadband
`services. Few subscribers to date take advantage of the limited satellite
`offerings, like DirecPC, now available. 41 But given that satellite broad-
`casting, or "DBS" service, is moving towards having 15 million subscrib-
`42
`ers,
`and that additional satellite systems have been licensed and are
`coming online, further offerings are likely in coming years.
`Although wireless technologies will likely become more important
`players in broadband transmission, at present they lag behind other tech-
`nologies. The most likely near-term solutions to the slow access speeds
`available to residential customers are those that make use of the landline
`telephone and cable networks. ISDN, DSL and cable modem services will
`thus likely see the fastest growth in the near future.
`4. Where the Residential Market Is--and Where It Needs To Be.
`Right now, residential broadband is more promise than reality. Al-
`though broadband access is now available in most states, coverage within
`those states is limited. A recent study found high-speed services offered to
`selected customers in only 10 percent of counties, although those counties
`43
`together contain 45 percent of the American population. This suggests
`that advanced services are starting to spread to residents of densely popu-
`lated areas, but rural customers will have a longer wait. Even if the prom-
`ises of telephone carriers and cable systems are met, fewer than half of
`American households will have broadband access in the next couple of
`
`38. See Annual Assessment of the Status of Competition, supra note 37, para. 83.
`39. See, e.g., Daniel Sweeney, LMDS: Finally Ready for Prime Time?, AMERICA'S
`NETWORK, Aug. 1, 1998, at 22.
`40. See, e.g., Howard Shelanski, Video Competition and the Public Interest Debate,
`in TELEPHONY, THE INTERNET, AND THE MEDIA 91, 100 (Jeffrey K. MacKie-Mason &
`David Waterman eds., 1998).
`41. See Les Freed & Frank J. Derfiler, Jr., Hughes Network Systems' Direct PC
`Internet access via satellite, PC MAGAZINE, Apr. 20, 1999, at 160.
`42. See Annual Assessment of the Status of Competition, supra note 37, para. 62.
`43. See State of the Internet: USIC's Report on Use and Threats in 1999 (visited
`Apr. 16, 1999) <http://www.usic.org/usic-state-of-net99.htm>.
`
`Netflix 1014 - Page 10
`
`

`
`BERKELEY TECHNOLOGY LAW JOURNAL
`
`[Vol. 14:721
`
`years. But substantial investment is being made in expanding such offer-
`ings, and greater availability is inevitable. The two unknown variables are
`price and speed of deployment.
`The best current prices for residential customers are, as indicated
`above, about $50 per month for a package of DSL or cable modem service
`and Internet access. Whether this will be considered affordable by the
`majority of Internet users is unclear. While the price seems high for those
`living on the median U.S. family income of about $30,000, other commu-
`nications services such as cable television subscription have proven to be
`fairly insensitive to income.44 Broadband access for Internet service might
`follow a similar pattern, especially if economically bundled with video
`and voice telephone service. But the economic structure of broadband de-
`mand is as yet unknown.
`It is very likely, however, that lower prices will substantially increase
`the spread of broadband subscribership. Indeed, future new purchasers of
`Internet access may be increasingly cost conscious. The available data in-
`dicate that average income of Internet users is declining. In 1995, the av-
`erage household income of an Internet user was over $50,000. 45 The latest
`Pew Center survey shows that the fastest growing groups of new Internet
`users are those with much lower income and educational levels than in the
`past.46 The survey finds that 23 percent of new users have annual house-
`hold incomes below $30,000 and that 39 percent of new users never at-
`tended college.47 This is a healthy development, but it also suggests that,
`over time, customers will be increasingly hard to attract at a given access
`price. So, for purposes of the growth of e-commerce, the price premium
`for speed will have to be low enough to reach customers farther down the
`demand curve for Internet access.
`
`44. See generally Robert Kieschnick & Bruce McCullough, Federal Communica-
`tions Commission, Do People Not Subscribe to Cable Television Because They Can Not
`Afford the Service? A Review of the Evidence (Aug. 1998) (unpublished manuscript, on
`file with author).
`45. See PROGRESSIVE POL'Y INST., THE NEW ECONOMY INDEX: UNDERSTANDING
`AMERICA'S ECONOMIC TRANSFORMATION 31 (1998).
`46. See Pew Research Center for the People and the Press, Online Newcomers More
`Middle-Brow, Less Work-Oriented: The Internet Audience Goes Ordinary (Jan. 14, 1999)
`<http://www.people-press.org/tech98sum.htm>; Bob Tedeschi, European Union Ad-
`(Apr. 27, 1999)
`vances E-Commerce Policies, N.Y. TIMES ON THE WEB
`(citing
`<http://www.nytimes.comlibrary/tech/99/04/cyber/articles/26commerce.html>
`Yankee Group estimate of $13 billion).
`47. See id.
`
`Netflix 1014 - Page 11
`
`

`
`1999]
`
`THE SPEED GAP
`
`From the perspective of electronic commerce, the challenge for the
`broadband market is to meet the growth targets announced by carriers, and
`to do so at prices that not only allow the carriers to make the required re-
`turn on investment, but also make broadband subscription attractive to a
`large number of households. As discussed below, the benefits to electronic
`commerce from such deployment are likely to be substantial for both buy-
`ers and sellers.
`
`II. THE IMPORTANCE OF BROADBAND CONNECTIONS FOR
`E-COMMERCE
`The convenience and novelty of online shopping has sparked rapid
`growth in the volume of electronic commerce. Recent estimates of retail
`sales over the Internet in the United States range from $8 billion to $13
`billion for 1998, up from $3 billion in 1997,48 and there seems little reason
`to believe the market will develop at a slower rate in the near future. Eight
`million Americans are estimated to have made online purchases this past
`holiday season.49 Established Internet businesses are becoming more user-
`friendly and sophisticated, while new entrants are coming (and going) at a
`rapid pace. "Infomediaries" that help consumers to search and sort online
`businesses have entered the market. And existing infrastructure is, for the
`moment, supporting substantial growth in the online marketplace. The real
`question is not whether there will be growth, but what trajectory it will
`follow. The ability of the telecommunications industry to provide fast and
`inexpensive pipes between online shoppers and Internet sites is an impor-
`tant factor in the answer.
`A. The Consumers' Perspective: Lowering Search Costs
`Many factors other than the cost and capacity of telecommunications
`connections limit consumers' demand for online transactions. Preferences
`for face-to-face interactions, privacy concerns about transmitting certain
`information electronically, and the inability to touch, try on, or tangibly
`compare certain products online constrain participation in electronic
`commerce-even among people who already use the Internet. Telecom-
`munications technology can contribute to easing those constraints, but is
`only one of several relevant factors. Network infrastructure is more cen-
`
`48. See, e.g., Sharon Linstedt, Santa Shops on the Web; $3.5 Billion in Online Sales
`Set This Season, THE BUFFALO NEws, Dec. 20, 1998, at 14C.
`49. See Mark E. Plotkin, How Traditional Companies Can Navigate the Web,
`LEGAL TIMEs, Mar. 1, 1998, at S32.
`
`Netflix 1014 - Page 12
`
`

`
`BERKELEY TECHNOLOGY LAW JOURNAL
`
`[Vol. 14:721
`
`trally relevant to the transaction costs of exchanges consumers do under-
`take electronically, as well as to the ability of online merchants to expand
`the range of transactions consumers are willing to engage in on the Inter-
`net.
`
`Basic, copper telephone links generally allow data to be retrieved at a
`rate of about 56 kbps (at best). At that speed, still images download slowly
`and video displays can take prohibitive amounts of time. For example, to
`download a 3.5 minute video clip through a standard 56 kbps modem
`takes more than 20 minutes.5 0 Even with a fast ISDN line, which transmits
`at about 128 kbps, that clip takes 10 minutes to retrieve.51 Such time re-
`quirements restrict the ability and incentive of potential customers to re-
`trieve useful or necessary product information and reduce the number of
`transactions for which they are willing to spend the necessary time on the
`Internet. Even if they are willing to retrieve slow-loading visual images
`from one or two sites, their ability to browse new sites and compare price
`and product offerings among online merchants is limited. Frustrated with
`the effort, some users will either buy from an established online seller, or
`buy the item at issue on their next trip to the (real) mall, as the world will
`not soon dispense with the necessity of some conventional shopping no
`matter how fast e-commerce expands.
`At faster speeds, consumers will obviously be able to explore more
`sites and, perhaps more importantly, to ob