Global "Last Mile" Performance Stats Going Public

Ookla, the company behind Speedtest.net, Pingtest.net, and the bandwidth testing apps deployed at many ISPs, has gone public with Net performance stats from 1.5 billion users (and counting). Their Net Index page displays download speed, upload speed, and connection "quality" from the EU and the G8, to countries, worldwide cities, and US states. Beginning today, the company is also making detailed (anonymized) data available to academics. "Ookla will also start surveying users about how much they pay for broadband and how much bandwidth they were promised by their ISPs. The results of those questions will go into building a Value Index, which will show how much people around the world pay per megabit-per-second for Internet access. In addition, by collecting postal codes from Speedtest users, Ookla hopes to map broadband service to local economic conditions, Apgar said. The Speedtest data could give the US government far more information to work with in setting priorities for its National Broadband Plan..."

Re:Moldova?

[Anonymous Coward]

Few have Internet there and the ones that do probably live in the capital city. It's easy to get a good index with one heavily wired city. It was only a few years ago that people in the villages didn't have inside plumbing. Considering the amount of people *leaving* those villages (and even the country), I doubt there is much reason to wire up the villages. Many villages do have cell access and the limitations that come with it.

Re:Disclaimer: I am an unabashed American.

[WrongSizeGlass]

The top 3 US States were:
*Delaware (15.56)
*Rhode Island (15.21)
*Massachusetts (15.01)

Bottom 3 US States:
*Montana (5.02)
*Idaho (4.29)
*Alaska (2.27)

NOT Speedtest.com, Pingtest.com

[pgn674]

The addresses are Speedtest.net [speedtest.net] and Pingtest.net [pingtest.net]. And yeah, I checked to make sure I got the capitalization correct.

speedtest.com is a squatter, and pingtest.com redirects to bandwidthplace.com, which looks awfully shady. Whois says it was registered by proxy, the Better Business Bureau has no record on that phone number, and neither does Google.

Re:The US looks pretty terrible.

[0racle]

Verizon FIOS tops at 50Mbit/20Mbit down/up for $139/month according to their site.

Now compare that to this [japantoday.com] from Japan.

"KDDI Corp will launch a fiber-optic communications service with upload and download speeds each of up to one gigabit per second on Oct 1. ... KDDI will charge 5,985 yen in basic monthly fees for Internet and telephone services, down 1,155 yen from the current price."

Yes, they said lowering the price. XE converts 5,985 yen to $66.29 USD. $66 for 1Gbit compared to $139 for 50Mbit.

In everything, from the bottom all the way to the top, American internet speeds and price absolutely suck.

Re:Hmmmm...

[corbettw]

Why should ISPs be held to a higher standard than automobile manufacturers, banks, insurance companies, the health care system, defense contractors, oil companies, mortgage brokers, Wall St financiers, and "family" farmers like ADM?

Re:Sometimes

[drachenstern]

Eh, my iPhone gets better speeds in [45 miles nearby large town] than it does on my DSL. #JustSayin ...

Re:Self-selection bias

[jmrives]

From what I can tell, the data they gathered is based on users around the world using their web site, Speedtest.net. So, there is no estimation on the part of the users. Also, the user who is testing their download and upload speeds does not have to be very tech savvy. All they have to do is open a browser, navigate to the site and click on a button to start the test.

Re:Self-selection bias

[talcite]

It also doesn't differentiate between mobile and fixed broadband speeds, which should affect the numbers significantly.

Why can't it differentiate between mobile and fixed broadband speeds? The user agent string from a mobile browser should be different from a desktop one. The only exception is if the mobile connection is tethered.

Re:The US looks pretty terrible.

[The_countess]

we'll probably just deny it and say that we are and whine about our lack of population density making it hard to build more infrastructure.

What part about that is untrue though?

People love to bring up Japan and South Korea and how fast their infrastructure is, but I don't see why it is not valid to bring up the disparities in size and population density.

South Korea is about the size of Kentucky with much higher population density and Japan is 90% of the size of California with roughly about %50 more population density.

Our Internet here is made up a number of competing telecoms and transit/peering agreements work great..... but when you have to keep putting fiber runs that are longer than the entire countries of South Korea and Japan why is it any big surprise that bandwidth costs more in the US?

sorry but those costs are only a tiny tiny fraction of the cost of internet connections. and south korea for example has nothing to link up to at all without expensive undersea cables linking it to Japan Europe and the US. your real problem is a lack of competition at the last mile. What you have a many regional mini-monopolies you have a choice of what? 2, maybe 3 ISP? on ADSL i have so many choices i dont even know them all, but i could name 8 big ones off the top of my head available nation wide. on cable you only have one (for the moment) but with the competition from ADSL they are kept reasonably honest. al thou now that they have speeds for up to 120/10 (close to 10 times whats achievable on average on ADSL) that could change of course. why do we have so many? because the government stepped in and forced network operators to allow 3de parties on their network at reasonable rates. result : we (the Netherlands) are position 8, without a large amount of fibre connections compare that to Belgium. slight smaller country, similar wealth and population density. they are next to the US on the ranking. their government didn't step in, their speeds are much slower and maybe more importantly they still have (strict) download limits which we haven't had, (not even for the cheapest connection) for years.

Re:The US looks pretty terrible.

[Solandri]

The vast majority of Canada is unpopulated or sparsely populated [statcan.gc.ca]. 90% of Canadians live in a 200 km strip along the U.S. border [bbc.co.uk]. Distance from Vancouver to Halifax is 4443 km, giving a 200 km strip an area of 888,600 sq km (which includes a lot of water, but ignore that). Canada's population is 33.2 million, 90% of that is 29.8 million. So 90% of Canadians live in a population density of 33.5 ppl / sq km. The U.S. has a population density of 32.1 ppl / sq km.

From net index site [netindex.com], the U.S. has an average connection speed of 10.16 Mbps. Canada has an average connection speed of 7.89 Mbps.