The meme of “America is falling behind in broadband” continues to persist, this time fueled by a recent report from Ookla’s speedtest.net which places America 33rd in average broadband speeds. The argument is deceptively simple: U.S. broadband appears to be slow, presumably because there aren’t a large number of broadband providers to choose from, therefore we must do something drastic to change the situation.
However, this line of thinking does not hold up to significant analysis and distracts from legitimate policy debates. To be sure, broadband is a critically important input to our economy, education, and daily lives, and it is easy to think of it as a modern-day utility. But setting up a utility-like regulatory regime would be a large step backwards, serving to further entrench incumbents and kill incentives to upgrade our networks. Such a move is totally unwarranted by current facts.
America is actually doing quite well when it comes to broadband speed. Akamai, a content delivery network, puts the United States as number 8, not 33, in the world in its latest State of the Internet report. Moreover, if you compare the speeds of U.S. states instead of the country as a whole (which is often a closer comparison in terms of size and population), U.S. states would sweep 10 of the top 15 spots. Why the large discrepancy between Ookla and Akamai measurements? One main reason is that Ookla utilizes a small number of servers to run their “speedtest” application, usually located in major cities, giving an advantage to smaller countries. Akamai measures the flow of traffic from their content delivery networks to end users and gives a much more controlled and accurate measurement. Moreover, the fact that Ookla’s results are self-initiated, usually when users are having trouble with their connection, also skews the sample.
Regardless of which ranking is closer to the truth, though the evidence suggests the Akamai rankings are, simple country comparisons are not a good guide for policymakers. One of the key differences obscured by these comparisons is population density. Countries like South Korea (487 people per square km2), Hong Kong (6,544/km2), and Japan (336/km2) are able to achieve fast networks because of the large numbers living in high-rise apartments. It is far, far easier and cheaper to provide high-speed broadband to a thousand people living in a single building than to a thousand people in U.S. suburbs, where population density averages 43/km2.
Furthermore, the United States differs from many other nations in that major broadband providers generally do not “cherry pick” densely populated urban areas to deploy to. Therefore, they have to charge the same price whether they serve low-cost, densely populated neighborhoods in cities or high-cost, less densely populated areas at the edge of metro areas. Ookla’s measurements are easily skewed by international networks that choose only to serve the most economical areas of town.
Many of these countries also had very little existing network infrastructure like cable TV and substandard copper infrastructure, making it easier to roll out a brand-new fiber network. Finally, these networks were often built with the help of large government subsidies, allowing these nations to make generational improvements in technology.
The U.S. model of intermodal competition in broadband, whereby different technological platforms compete with one another, has served us well. It leverages our existing infrastructure while preserving network operators’ incentive to invest in the networks that serve our country. It leaves open the possibility of new entry by a yet-unknown technology and avoids ossifying a particular platform as our only option. Ookla’s skewed measurements putting us at 33rd are not only inaccurate, but also do nothing to justify a departure from this track.
–Doug Brake is a Telecom Policy Analyst with the Information Technology and Innovation Foundation.