Anatomy of a deceptive broadband chart

WorldComp by Shutterstock

The debate over international broadband comparisons goes on. There’s lots of evidence the U.S. is doing well. Those who want more government control of the Internet, however, aren’t always happy about this success. Good news makes a big change in policy — from a relatively unregulated market to one of their preferred big government models — less likely. Thus the increasingly creative attempts to characterize the U.S. as a broadband backwater.

The chart below is a perfect example. I encountered it last week on Twitter, when across my stream came a tweet lamenting the fact that the U.S. trails France in broadband. I replied with a few links to actual international comparisons in which real data showed relative U.S. health. Back and forth we went, until it was time for the tweeter to play his trump card. Boom! A dispositive chart — showing  Americans pay more and get less than the rest of the world. 

Slovenians, according to the chart, get 100 megabits per second (Mbps) and pay just $37 a month. In the Slovak Republic, broadband users enjoy 60 Mbps for $28. Americans, meanwhile, get the shaft, paying $90 for 50 Mbps. On it goes. The unmistakable takeaway from a glance at the chart is that the U.S. trails the world.

Screen Shot 2014-03-03 at 4 39 47 PM (2)

Except the chart doesn’t say that at all. It took a lot of work to create such a misleading chart. We’ve annotated the chart (in red and bright blue below) showing specific problems, but we ran out of room. So let’s list the chart’s deficiencies here.

wrong broadband chart 1.0

  1. The title, “Countries with high-speed broadband,” isn’t false, but it excludes a lot of relevant information. Most nations with the fastest broadband networks, for example, are not on this list. A more descriptive title might have been, “Selected high-speed broadband plans in some countries.”
  2. One of the key metrics of the chart is “Megabits per second downloads.” But the chart doesn’t show actual download speeds. It shows advertised download speeds. There’s a big difference. Actual speeds in the U.S. are closer to advertised speeds than in most nations.
  3. The “megabits per second downloads” metric merely shows the existence of an advertised product, regardless of how many people have access to it or actually purchase it.
  4. For example, 100 Mbps is not at all a typical offering or consumer experience in France or Slovenia.
  5. Just three nations on this list are in the top ten nations in terms of actual average download speeds (South Korea, Japan, U.S).
  6. Just three nations on this list are in the top ten nations in terms of percentage of broadband connections delivering actual speeds above 10 Mbps (South Korea, Japan, U.S).
  7. The chart thus is not representative of the real experiences of the people in the nations it lists.
  8. Comparing broadband in large, low-density nations and small, high-density nations has been a fundamental problem over the years. This chart repeats that endemic failure. It compares large nations with relatively low-density populations (U.S. and Canada) with small and/or densely populated nations (South Korea, Estonia, Japan, UK). It is far more expensive to build broadband networks in large, low-density nations than small, high-density nations. Consider the cost of laying fiber to farms sprinkled across Nebraska, compared to wiring thousands of units in a single high rise apartment building in Seoul. Wiring the apartment building is far more economical. This is why in our analyses, we like to compare U.S. states to these small nations. Still not perfect, but better.
  9. The chart cites “flagship” products in some nations, but not others. For example, the Solvenian “100 Mbps for $38” might be the top advertised product in the nation. But the chart does not cite Google Fiber, an American product which advertises 1,000 Mbps for $70. The Google Fiber bar would have extended off the chart. It’s true, Google Fiber is not a good representation of the typical U.S. experience. But that proves the point. The chart does not even cite other widely available American products that have a more attractive Mbps-per-dollar ratio than the one it does cite.
  10. The chart does not address the vast bulk of broadband plans, which aren’t at the very top-end of the speed and price spectrum but which nonetheless determine the experience of the majority of a nation’s Internet users. The assertion that the U.S. has the most expensive top-end broadband plans can be misleading because the U.S. also has the most inexpensive entry-level broadband plans in the OECD. Is it a problem we, in effect, let our high-income power-users pay higher prices in order to, subsidize lower-income email checkers? Perhaps that’s not a bug but a feature.
  11. The chart uses Purchasing Power Parity (PPP) for its price comparison. PPP can be useful, but it can also be misleading. Like other price indices, it is imperfect. Also: nations like Switzerland, Sweden, Norway, and Denmark would not have looked as good on this chart — their PPP comparisons don’t reinforce the “high price U.S.” argument — so they weren’t included.
  12. The chart does not include information on adoption rates. If lots of consumers in nation A subscribe to plan A1 (45 Mbps for $60), and very few consumers in nation B subscribe to plan B1 (“100 Mbps” for “$30”), which nation is better off?
  13. The chart does not, I don’t believe, include taxes and fees, which can be substantial, especially in some European nations, and which can thus significantly alter the total cost to consumers.

There is no perfect comparison. But if we use better data and focus on what really matters, we can build more useful analyses. My colleagues Richard Bennett, Roslyn Layton, and Gus Hurwitz have all contributed useful perspectives lately.

And I’ve made attempts as well. Akamai’s reports of actual broadband speeds, for example, show that many U.S. states enjoy some of the fastest broadband in the world and that the U.S. as a whole enjoys the fastest speeds of any non-small nation. 

Several years ago, I developed a simple but powerful metric that tends to encapsulate many relevant factors like broadband penetration, speed, price, and even usefulness and availability of content and apps. That metric is network traffic per capita and per user. When I began tracking this measure, the U.S. generated and consumed around 60% more traffic (per capita and per user) than Western Europe. Only South Korea exceeded the U.S., and Canada and the U.S. were about even. In the last couple years, however, the U.S. lead has grown, and the U.S. now generates and consumes three times the per capita/user data of Western Europe, more than twice that of France, and has nearly even caught up with longtime outlier South Korea. The U.S. could not generate so much more traffic if its networks were slow, not widely deployed, not widely subscribed, if high prices discouraged subscription and usage, or if our digital value chain was not producing useful services, content, and apps.

This isn’t the last word. But it’s at least a reminder to be wary of claims of “the U.S. broadband problem in one chart.”

One Response to "Anatomy of a deceptive broadband chart"

Leave a reply