Comment the plural of "anecdote" (Score 2, Interesting) 233
first a nugget of fact, then some commentary:
1. When we moved to Portland, Oregon, we had Qwest come out to the house to rewire one of the phone jacks because the mooks who hooked it up to the outside world crosswired the connections- we didn't even have dial tone. After the tech fixed the problem, first thing he did after confirming DSL sync was to run a speed test. I asked him if that was SOP and he said that he was trained to always run a speed test for new customers- he suggested that it might be part of an upsell but that he doesn't like selling so he never comments (oh, you're only getting 750k down, but you're in an area where 7/1 MB service is available... did you know you can upgrade for just $3.50/month!????
2. I agree with earlier commenters- there is probably a self-selecting sampling bias.
3. Because of #2, any "data" they collect is probably very skewed towards computer-savvy users who are demanding higher-speed services and using their website to check if the service they're getting matches what they're paying for. Unless there are some details of the methodology that they're not telling us about, the survey probably reports higher bandwidth than actually is delivered to the majority of people with net access in those cities. If it's just a simple aggregation & average of whoever decides to click on speedtest.com from inside a given city's IP range, well, that probably tells you something... but it's probably not a good proxy for a complete picture of "last mile" connectivity.