For example, many of the charts and comparisons I’ve seen, including the one in the USA Today article cited above, appear to compare actual delivered U.S. speeds with advertised speeds in other countries.  For example, in delving down into the Japanese numbers, which is often cited as among the fastest countries in world when it comes to broadband speeds, you find that roughly half of all Japanese broadband connections are not fiber but rather DSL.  While loop lengths are shorter in Japan, the reality is that DSL speeds are far lower than the 63.6 mbps cited in the USA chart.  That fact alone suggests that many of these comparisons are not accurate.  .

Second, the Internet is a network of networks and speeds measured across the net can vary greatly by time of day, the number of router hops involved, the type of network involved, and even factors such as major sporting events that may be occurring (lots of people can be online during a major sporting event, all at the same time, sharing local connections).     Measures based on local routers can be more accurate but even in these cases, downstream and upstream are often very important in today’s online world and you seldom see upstream speeds measured. 

Many reports assessing broadband speeds used voluntary tests to generate results.  Web sites and servers connected to those sites are advertised and consumers voluntarily take the speed tests which generate average speeds based on those tests.  Obviously these tests can vary a lot depending on the number of people using each specific technology (i.e., cable modem, DSL, WiFi and so on) who decide to log on and take them.   They can provide some useful data but again, they often don’t measure upstream speeds at all nor are they really as accurate as weighted surveys might be. 

Ironically, one of those testing sites, Speedtest.net, does have servers in many countries and it does publish its own reports based on speed tests from citizens around the world.   They publish periodic reports about their results and the 2007-2008 report found that average delivered speeds put the U. S. in the upper tier of countries world wide when it comes to speed, near countries like France which is often cited for its high speed DSL service.   Now there are many reasons for this, including the fact that while DSL in the cities in France is very fast, speeds in more rural areas where copper loops are longer may be much slower.   This would tend to make the French averages lower. 

I’m not trying to suggest here that I have the answer as to who is fastest  The point is it depends on a range of factors but more importantly using advertised speeds to compare countries is fraught with problems.  Japan’s average speed of 14 megs as reported in Speedtest.net is far below reports about available advertised speeds in that country.  Further, the Japanese themselves admit that even though they have fiber deployed to many homes, the congestion problems they face are severe.  So delivered speeds may be suffering there, even though they have new technologies like fiber to the home. 

Pricing is another area where the U. S. frequently gets criticized.  Many advocates say U. S. broadband prices are far too high and cite statistics based on a price per megabit.  Yet, again the statistics cited are not quite that straightforward. For example, some say that Japan has one of the cheapest broadband offerings on a per megabit basis in the world.  Yet, the reality is more complex.  Most reports say Japan’s broadband service only costs about $30 a month for tens of megabits of service.  Yet, the reality is that Japan’s service is priced differently.  You have to subscribe to two different services to get an equivalent product to that offered in the US.  For example, not only would you subscribe for the network connection (essentially the equivalent of getting the physical line), but you also have to subscribe to an ISP (the network in Japan is divorced from the content).  For access to a 100 megabit connection and Yahoo! ISP service, you’ll be forking out between $55 and $60.  So both because the service is more costly than many say and because the average throughput delivered is probably far below what is advertised, the comparisons often made between U. S. broadband prices and those of overseas countries are flawed and inaccurate. 

What is even more frustrating in all of these debates about statistics is that if you look to the future, the U. S. is moving to next generation technologies faster than just about any country and we are making progress.  Gartner just came out with a study that suggests that by 2012, the U. S. will be among the top five countries in the world in terms of broadband penetration to the home.  In fact, we will be very close to those countries above us including the Netherlands, and tied with Japan.   New technologies, like wireless LTE broadband technologies which Verizon will soon be deploying, are cited by Gartner as a key reason for the 20 percent jump in U. S. broadband penetration.

The statistics cited regarding broadband speed, penetration and pricing are confusing, often compare apples and oranges, and in most cases don’t measure really important factors such as who is deploying next generation technologies most rapidly.   Mark Twain had a very earthy saying about statistics – “There are lies, damn lies, and statistics”.  He meant this as a humorous observation about how easy it is to assume numbers are always right.  But it is not the numbers per se but rather how they are used and how comparisons are made that is key.