October 4, 2001
Videoconferencing May Get Much-Needed Critical Mass
By HAL R. VARIAN
ill the current reluctance to travel stimulate widespread use of videoconferencing? The stock market seems to think so. Shares of companies that specialize in videoconferencing have shot up 30 percent or more in the last two weeks.
But people have been forecasting the takeoff of videoconferencing for decades. In the early 1970's AT&T (news/quote) offered Picturephone service in Chicago for $86.50 a month. Jeff Rohlfs, a Bell Labs economist who was involved with the project, describes the history of this technology in his new book, "Bandwagon Effects in High-Technology Industries" (MIT Press).
Based on his experience with the Picturephone introduction in the early 1970's, Mr. Rohlfs came up with a concept he called network externalities. According to Mr. Rohlfs's definition, a good or service exhibits network externalities when its value to a potential purchaser depends on how many other people have purchased it.
Fax machines and Picturephones are standard examples. The fax machine was a big success, while the Picturephone was a failure. Why the difference?
Mr. Rohlfs developed a model that showed that "network goods" would be successful only when they were able to achieve an initial critical mass of early users. If the initial number of purchasers exceeds this threshold, the market takes off. Otherwise, it falls back to failure.
The Picturephone had 200 users at its peak, apparently far less than critical mass, and the product was withdrawn from the market.
Nobody cares about a failed technology, and Mr. Rohlfs's work was rarely cited in the decade after its publication. Then in the 80's and 90's we saw fax machines, video players, the Internet, e- mail, and the Web, all of which exhibited strong network effects.
Other economists, like Michael Katz and Carl Shapiro, picked up on the idea and studied strategies that companies could follow to achieve critical mass. Bob Metcalfe, inventor of the ethernet, independently promoted a related concept now known as Metcalfe's Law: the value of a network grows as the square of the number of users.
Talk about network effects exploded in the late 1990's. It is one of those rare concepts that you could explain to budding chief executives in five minutes and they could pitch to venture capitalists for six months — or until they got financing, whichever came first.
Videoconferencing never achieved critical mass because it has always been more attractive to hop on a plane. Real-time face-to-face meetings are costly, but that's the way business has been done for centuries, and it is hard to change ingrained patterns of behavior.
Those companies that have experimented with videoconferencing have had mixed success. Low-cost videoconferencing is flaky enough that you can't really count on its working, and the high-quality technology is still pretty expensive.
But the days of blithely hopping on planes are over, at least for a while, so many companies will experiment with the technology in the coming months. As more companies use it, the technology will improve and videoconferencing capability will become more widespread. Perhaps it will finally achieve the critical mass that has eluded it for so many decades.
Most large American companies have a good enough network infrastructure to experiment with videoconferencing. The situation is different for residential broadband, since only about 6 percent of American households have high-speed network connections. Silicon Valley would love to see residential broadband take off, since it hopes that would stimulate demand for PC's and related consumer electronics. This week there are two meetings in Washington devoted to this topic, part of a series of efforts by technology and telecommunications companies to change regulatory constraints that they say inhibit residential broadband deployment.
Two years ago, I worked as part of a team to estimate residential demand for broadband. We gave about 80 university faculty members, staff and students residential high-speed connections. Our setup allowed users to change their bandwidth instantaneously, and we could charge them based on their choices. We changed prices for bandwidth every week and observed how people responded.
Our findings were summarized in several reports, but they boiled down to the following. First, the average user isn't willing to pay much for bandwidth, at least given the applications available in 1998-99. Second, those willing to pay for bandwidth were primarily technical and administrative workers, who appeared to use it for telecommuting. Third, most people were willing to pay a significant premium for a flat monthly rate, rather than a rate that varied by use.
The General Accounting Office reported results of a survey run in late 2000 that indicated that about 50 percent of Internet users in this country reported that broadband was available to them, but only 12 percent actually signed up for it. This is consistent with our first finding: people like broadband, but they aren't willing to pay a big premium for it.
Dial-up Internet access is widely available for around $20 a month, while broadband costs about $50 or more a month. It just isn't worth it for many potential customers.
The people who really like broadband are technically oriented telecommuters. Uploading and downloading files is a lot faster, and broadband makes working at home a lot easier. This suggests that broadband providers might well consider offering service packages oriented to home workers.
Maybe videoconferencing will be part of the mix, if it becomes widely used in business.
Noncommercial use of high bandwidth is more problematic. Opinions are mixed about how much extra people are willing to pay for movies on demand, especially when there are so many movies available via cable and local video stores. There seems to be some sort of market for music, but worries about piracy are likely to slow the availability of music on the Internet.
Residential broadband will come, but it will be slow in arriving until compelling applications are available.