[saraid216]

I was curious, so I pulled their appendix (p39) to actually look:

Top Twenty Percent above 10 Mbps:

  irb(main):057:0> bar.sort { |a, b| a[:perc_above_10].to_f <=> b[:perc_above_10].to_f }.reverse.collect { |item| item[:country] }[0..20]
  => ["South Korea", "Japan", "Netherlands", "Switzerland", "Czech Republic", "Hong Kong", "Belgium", "United States", "Denmark", "Sweden", "United Kingdom", "Finland", "Canada", "Ireland", "Norway", "Austria", "Taiwan", "Israel", "Russia", "Singapore", "Poland"]

Top Twenty Percent above 4 Mbps:

  irb(main):058:0> bar.sort { |a, b| a[:perc_above_4].to_f <=> b[:perc_above_4].to_f }.reverse.collect { |item| item[:country] }[0..20]
  => ["South Korea", "Switzerland", "Netherlands", "Czech Republic", "Japan", "Israel", "Denmark", "Canada", "Austria", "Hong Kong", "United Kingdom", "Belgium", "Romania", "Luxembourg", "United States", "Germany", "Russia", "Sweden", "Poland", "Finland", "Spain"]

Top Twenty Average Speed:

  irb(main):059:0> bar.sort { |a, b| a[:avg_speed].to_f <=> b[:avg_speed].to_f }.reverse.collect { |item| item[:country] }[0..20]
  => ["South Korea", "Japan", "Netherlands", "Hong Kong", "Switzerland", "Czech Republic", "Sweden", "Ireland", "United States", "Belgium", "Denmark", "United Kingdom", "Finland", "Austria", "Canada", "Norway", "Taiwan", "Israel", "Singapore", "Germany", "Poland"]

If you intersect all three for "higher than United States", you get:

  irb(main):080:0> (ary_one[0...ary_one.index('United States')] & ary_two[0...ary_two.index('United States')]) & ary_three[0...ary_three.index('United States')]
  => ["South Korea", "Japan", "Netherlands", "Switzerland", "Czech Republic", "Hong Kong"]

What bothers me about the data is the implicit distribution skews in the Peak and Avg numbers, but that's a problem with every line, so not material to the subject here. It may signal a methodology weirdness, but it's lunchtime and I don't feel like paging back to see if I can find something showing that Akamai is counting things it shouldn't be.

[adventured]

I always find it fascinating when people take the data and proclaim that America should be stacked up against nations with single digit millions of people living there, much less postage stamp size areas like Hong Kong or Luxembourg.

Most of the comparisons are hilariously ridiculous. The proper comparisons are: China, Brazil, Japan, Indonesia, UK, Russia, India, Germany, France, Turkey, Italy, Mexico - large scale industrialized nations.

Comparing Norway or Luxembourg or Hong Kong with the US, as though they can possibly be compared in any sane respect what-so-ever is absurdity to the ultimate degree.

The fact that we compete at all against tiny nations with 5 or 8 million people is a stunning accomplishment.

[rayiner]

Half the population of South Korea lives in the Seoul metro area, which has the same population density as the city of San Francisco. The whole of Hong Kong is a single city, again about as dense as San Francisco. The whole country of Japan is about as dense as the state of Massachusetts, the 2nd densest U.S. state. It's about twice as dense as New York, the 7th densest U.S. state.

Dense U.S. states outperform the overall U.S. average significantly (and comparably to Japan): http://www.bizjournals.com/boston/blog/techflash/2014/04/rep....

Out of the 10 U.S. states with the fastest internet, 8 are just the densest states (+D.C.) in a different order.

[adamnemecek]

The thing is that the internet speeds even in the dense areas are disproportionately slower.

[rayiner]

How do you get that? Massachusetts's average is 13.5 Mbps. Japan is 12.8. New Jersey, Maryland, and Delaware are 12 and up, above Switzerland, Sweden, Czech Republic, etc. The dense U.S. states compare very favorably to the top-5.

South Korea is the outlier because half the country lives in a single megacity with the density of San Francisco. The U.S. can't compete with that geographic advantage.