This is not a discussion on the beautiful art of counting without counting (sorry), but instead my soapbox to project questions about the act of counting in real life. I first considered the question when I realised one day that I had no idea what the answer was to “How many people were there at critical mass yesterday?” (Clearly, going by the date of this post, it is not a recent question).
The same sort of difficulty (and perhaps moreso) is present when asking how many people showed up at a protest (for example, though we could also say pillow fight). Crowd estimate to me seems like something you only get accurate or consistent answers for at modern sporting stadia, where every admitted person has to go through a turnstyle.
I can think of two potential techniques (though lack the resources or inclination to test these out). For a moving crowd, one could try to get an accurate count by counting numbers as they crossed a particular point along the route (probably most practical afterwards with a videotape, though the time involved may not be considered worth it). The other I can think of is to gt a desnity of people estimate, then multiply by the total area taken up. Edge effects are important here, since it is both hard to define precisely the edge of a crowd, and presumably person density is not uniform, but tapers off towards the edge.
Anyone know what is actually done in practice? Or do people just make up wild guesses.
Ceci n'est pas un blog 