Getting better estimates of UFPJ's march
There’s been much grumbling among the anti-war protestors who attended the January 27th march that the press got the numbers wrong. The Associated Press report stated “tens of thousands,” apparently because an unnamed police source suggested he thought (without having done any sort of systematic estimate) there were less than 100,000 people. United for Peace and Justice didn’t help things by claiming to have 500,000, which (having been there) was simply not credible.
As I scientist, this deeply frustrates me. This is an easy problem to solve. UFPJ needs to negotiate for a digital camera to be placed at a high location with a view of the rally site (the Washington Monument would have been perfect in this case). The camera would take a photo every 30 minutes or so, and the picture would be uploaded to the Internet. From there, someone (and I’d be happy to help) would download the picture, geometrically rectify it (this accounts for the slant of the picture, the fact that it’s not straight up and down), and then make an estimate of the crowd size (one typically counts the density of the crowd in two or three specific 100 m2 sections of the photo, then multiplies by the area of the crowd). The estimate, as well as the picture (archival and geometrically rectified) it’s based on, could then be placed on a website, in near real time. Journalists would be likely to report the estimate, as long as it was clear that the group doing the estimate was not directly affiliated with the march, and as long as they could double-check the calculation with the photos. The whole thing could probably be done for a couple thousand dollars worth of fees, rental equipment, etc., and would provide credible estimates to the press.
