Greetings. A week ago in Why I'm Skeptical of the FCC's Call for User Broadband Testing, I questioned the FCC's methodology in loudly promoting a pair of Java-based broadband speed tests on their Web site, in close conjunction with the release of their National Broadband Plan. In that posting, I discussed a number of technical reasons why I was concerned that the FCC tests would confuse consumers and generate data of very limited value.

Now comes word from the FCC that they actually do not even endorse these tests! That amusing pronouncement comes in a Los Angeles Times article discussing how a reporter contacted the Commission after finding wildly varying results in his own use of the FCC-sponsored tests, and received finger-pointing responses from the tests' vendors -- each claiming that they were accurate and that the other test must be screwing up. One vendor even blamed the M-Lab servers.

The FCC also noted that "there might be more confusion for consumers than we realized" (technical term: "Duh") and promised to add new disclaimers to the FCC site regarding "widely varying results" from using the tests.

It's now more apparent than ever that this particular FCC effort at consumer broadband testing was in reality likely little more than a publicity stunt. Given the many variables faced by these sorts of manually-triggered "one-shot" consumer tests, they are by definition usually of use only in the sense of rather gross measurements. That doesn't make them worthless by any means, but does limit their applicability, and as we've seen can exacerbate consumer confusion, especially when vendors apparently don't always want to mention up front that these sorts of limitations exist.

I'm glad that the FCC now plans to better disclaim the constraints of these broadband tests. Of course, they should have done this originally, or more appropriately not have so loudly promoted these tests to the media with unsupportable hoopla in the first place.

--Lauren--