Take the U.S. broadband census, wherever you are

One of the email discussion lists I subscribe to has seen some coverage of this site, which claims to be collecting data on US broadband speeds, to what end I know not. Based on what I've seen, I only hope that the data they harvest is not employed in any sort of lobbying or policy-making initiatives. Here's why.

Initially it makes some credible-sounding statements about the role of broadband in American society:

"More and more Americans depend on high-speed internet service for education, commerce and entertainment. Broadband is the gateway to the information superhighway."

Hey, 10 bonus points for using "information superhighway" from the get-go!

It continues:

"BroadbandCensus.com is dedicated to providing the most comprehensive public and transparent collection of data about local broadband speeds, prices, availability, reliability and competition. You can help us fill the broadband data gap by Taking the Broadband Census."

Sounds good, happy to help. How else can I get involved? Maybe I should join the research committee, after all:

"The Research Committee will help the Broadband Census' efforts to map out broadband availability, speed, competition and price in an empirically sound fashion."

I like what I'm hearing. So, feeling patriotic, and in the spirit of courteous driving on the information superhighway, I took the test - twice, once claiming to be a Comcast customer, once as an AT&T customer (selecting "fiber" just for laughs). Here are my results:

Thank you for taking the Broadband Census. Your input is appreciated. It will help educate broadband consumers all over the country.

Promised Downstream Speeds: NA

Actual Downstream Speeds: 3.75299 Mbps

Promised Upstream Speeds: NA

Actual Upstream Speeds: 0.444 Mbps

Go to your ZIP code: 38117

Go to your provider's page: Comcast

Promised Downstream Speeds: NA

Actual Downstream Speeds: 4.33679 Mbps

Promised Upstream Speeds: NA

Actual Upstream Speeds: 0.476 Mbps

Go to your ZIP code: 38117

Go to your provider's page: AT&T

It looks like the data might go straight onto the site with no mediation. I checked out the 38117 ZIP code page after my test, and there was only one result from an AT&T user, who had given the service four stars - the rating I gave it in my second test. I have to assume this was my response. I could repeat the test to confirm, but I'm getting bored now.

There are huge problems here. I am in the UK, which is where I took part in the U.S. "census". As the speed test requires no identity assertion, and clearly does not exclude non-US IP addresses from taking part, I would assume that anyone can submit as many bogus entries as they want to, from anywhere in the world. Not that I would ever suggest or endorse such behavior...

UPDATE: Despite what I initially wrote and pathetic as it may seem, I actually did subsequently go back and take the test for a third time, claiming to be an AT&T subscriber again, but this time giving the service only one star. At this writing, the 38117 ZIP code page contains only two ratings for AT&T - a four star, and a one star, leading me to the inescapable conclusion that both of these results were generated by my bogus entries, taken at face value, despite coming from a legacy ntl.com IP address, which various free analytics tools clearly identify as being based in London. A for intentions, F for execution.