South Korea has the fastest home Internet service in the world, with a downstream speed 100 times faster than the average in Sudan, the operator of Speedtest.net reported on a new Web site that compiles results from consumers running its online test.
When South Koreans tested their wired broadband connections over the past 30 days, they found an average downstream speed of 34.14M bps (bits per second), according to the Net Index, which was introduced on Tuesday by Ookla, the creator of Speedtest. That was several times the worldwide average of 7.67M bps and 100 times as fast as the 340K bps downstream speed in Sudan, the lowest average out of 152 ranked countries.
Latvia (24.29M bps), the Republic of Moldova (21.37M bps), Japan (20.29M bps) and Sweden (19.78M bps) rounded out the top five countries for downstream broadband. The U.S. was ranked 26th in the world, with an average downstream speed of 10.16M bps. Upstream rankings were similar, with South Korea leading at 18.04M bps and the U.S. in 27th place with 2.21M bps. The world average was 2.10M bps.
Residents of Seoul, the world's fastest major city for broadband, enjoyed just slightly higher bandwidth than the South Korean average, registering 34.66M bps. Riga, Latvia, came in second with 27.90M bps. San Jose, California, in the heart of Silicon Valley, was the fastest U.S. city, with an average of 15.03M bps downstream.
Ookla, which is based in Seattle, also ranked the states and major cities within the U.S. Delaware had the fastest average downstream speed, at 15.58M bps, while Alaska had the slowest, 2.77M bps.
The numbers are filtered in certain ways to prevent distortion one way or the other, according to Ookla co-founder and CEO Mike Apgar. The statistics only include countries where tests were generated from at least 75,000 unique IP addresses. Some, including South Korea's neighbor North Korea, did not make the list of 152. For Ookla's formal Top Ten, only countries with at least 100,000 unique IP addresses used in tests are qualified.
Ookla is making the data available to help consumers, ISPs (Internet service providers) and regulators make more informed decisions, Apgar said. Previously, the company has provided raw data to researchers and companies through one-on-one arrangements. It also provides some data to ISPs, which can present Ookla's broadband-testing tools to their subscribers under their own brands.
Though the speeds may appear high for nationwide or citywide averages, Ookla says Speedtest detects more than just the bandwidth that a single PC browser can consume. It uses multiple threads of data to determine the full capacity of an Internet connection, which could be used by several devices around a home, Apgar said.
At least 90 percent of the results on which the index is based came from wired, residential broadband connections, which is the focus of the index, Apgar said. Ookla can detect a cellular session by the name of the ISP and manually filters out the results of tests that were probably conducted from a business, he said. Test results from Clearwire's WiMax wireless broadband service are included.
How much bandwidth consumers can actually get is a heated topic in the U.S., where various factions are debating how best to make more and better broadband available. The U.S. Federal Communications Commission has had a program to encourage consumers to measure and report their Internet connection speeds, using tools including Speedtest.
Free Press, an activist group on broadband issues, commended Ookla for making the data public but called it unscientific. Most importantly, Speedtest results come from users who know about the service and are interested enough in their broadband performance to test it, said S. Derek Turner, research director at Free Press.
"It's data, but it's self-selected data, and so it's not really useful for the purposes of policymaking," Turner said. "Anecdotes matter in the public policy debate, but when you can get actual empirical data, policymakers should strive for that."
Though consumers run Speedtest by choice, the size of Ookla's sample should make it fairly representative, Apgar said. About 60 percent of the people running tests each month are first-time users of Speedtest, so it's not the same small group of users testing their lines over and over, he said.
"It can't serve as the only way to determine what the state of broadband is in the U.S. .... (but) it would be foolish to overlook a resource like this," Apgar said.