Why the Internet won't be metered

Point-to-point and flat rates are king

John Levine
Author and expert

September 2000

1. Introduction

2. Content isn't king — it's the court jester

Recent common wisdom has held that "content is king," meaning that the future will be dominated by Web sites that feed text, animation, sound, video, smell-o-vision, and 3-D feelies to individual users. Whereas it's quite likely that the trade press will continue to talk about content, it's extremely unlikely that content will be where the money is. Why not? Because customers are willing to pay more for one-to-one connection than for content.

Individuals and businesses already spend much more on point-to-point communication (phone and mail) than consumers spend on content and entertainment (music and movies). Odlyzko shows that the phone industry is not only bigger, it's growing faster than "content" industries. Movie theaters sell nearly $10 billion of tickets a year, but that's only two weeks' revenue for the phone industry. Similarly, the revenue of the recorded music industry is dwarfed by that of Fed Ex and the U.S. Postal Service. Even meterists such as Metcalfe believe point-to-point communication is the most likely type of traffic to be metered because that is what customers are willing to pay for.

E-mail has been the Internet's "killer app" for 30 years, and despite the hype, the Web hasn't changed that. If you had to do without e-mail or the Web, which would you forego? Everyone I know would give up the Web.

Odlyzko analyzes the content issue in much more detail, looking at the convergence of formerly unrelated businesses (for example, how the Internet destroyed the business of the Encyclopedia Britannica) and the way that Net-based technologies like Napster are rapidly making much traditional content delivery irrelevant.

3. How do Net charges work?

Regardless of how people use the Net, someone has to pay for the modems, phone lines, servers, routers, and data centers and the staff to run them. One can make a strong economic argument for "price discrimination," charging higher prices to people who are willing to pay more.

Odlyzko cites a hypothetical case of two people at the end of a long road who want Internet service. One is willing to pay $10,000, the other is willing to pay $20,000, and running the line out to them would cost $25,000. If the vendor charges $10,000 to each, both will sign up, but the revenue will be only $20,000 (not enough to run the line). If the vendor charges $20,000 to each, only the second will sign up, which still won't cover the $25,000 cost. But if the vendor creates an artificial distinction and offers high-speed service for $18,000, and low-speed service for $8,000, both customers will sign up, the revenue will be $26,000 (enough for a $1,000 profit), and both customers will have service for less than they were willing to pay. This makes economic sense even if the cost of providing the service to both is the same, although, ironically, consumers tend to feel cheated if they find out.

Price discrimination works for relatively infrequently purchased high-priced items, such as airline tickets, because people are willing to negotiate for expensive purchases. But it doesn't work for low-priced items where maintaining the price discrimination is more trouble than it's worth. Most ISPs offer flat rate plans because the market prefers them and it avoids customer resentment. (Odlyzko cites a woman who vehemently accused AOL of cheating her by charging her hourly, even though she paid less than she would under their flat rate plan due to her low usage.)

4. Users love low flat rates

In North America, almost without exception, residential telephone users can get flat rate plans with free local calls. (Internet growth has been slower in Europe where local calls are usually metered.) With great regularity, users pick flat rate plans even when their usage history shows that a metered plan would cost less. The reasons for this seemingly irrational preference appear to be twofold. By choosing a flat rate plan, users avoid an unexpectedly high bill if they make a lot of calls one month. Furthermore, most consumers don't want the hassle factor of deciding whether each phone call is likely to enrich their lives by 10 cents per minute.

Odlyzko examines wired and wireless telephone service pricing at length. Wireless service is inherently more expensive than wired, because one can add extra fiber capacity without limit, whereas the available radio spectrum is limited. Even when true flat-rate pricing isn't possible, the market still tends toward pricing plans that provide most of the benefits of flat rates. For example, wireless services offer purchase plans where the user buys as much service as they expect to use in a month, with service beyond the prepaid amount either being blocked or charged at a higher rate. Prepaid plans are good for providers because users tend to buy more than they plan to use to avoid surprise and hassle.

Flat rate pricing is more difficult when there is a large disparity in customers' usage patterns, but if the pool of users is large enough, an appropriate flat rate model can be determined. For example, ISPs can predict the minimum number of modems needed to ensure access for all users, even at peak usage times (around 8:00 PM weeknights). The peak number of phone lines and modems needed largely determines the ISP's costs, which are the same whether all or none of the available modems are in use at any given time. As with the phone example, many users are willing to pay more for flat rate or bundled plans than their metered usage would cost. Therefore, a flat rate structure can ensure a profit even if a minority of users, such as "campers" who stay connected day and night, get "more than what they paid for." When necessary, ISPs can discourage camping via monthly caps, limits on session length, or limits on peak time usage that are ample for normal users but too stingy for campers. A metered pricing plan wouldn't discourage peak usage, but would instead discourage use at off-peak hours with no cost savings to the ISP. Furthermore, metered pricing would most likely lower the revenue from the majority of customers with light usage patterns who would choose a flat rate plan if given the option.

Odlyzko's paper has much more than I can summarize here, and it's well worth reading all 150 pages, but its important messages are simple: People are willing to pay much more for point-to-point communication than for broadcast content, and successful services almost always move from complex pricing to low flat rates. As retail users move to DSL and cable connections, where each user pays for their dedicated connection, the pricing is invariably flat rate. The Internet may "change everything," but it's pricing structure is following a path laid down over the last 100 years.

5. Resources

6. About the author