Now that would be giving snake - so in the end it would be a poll about those who applies avatars, and once there the poll would as usual probably have shitty results - and do nothing.
Of course it does do away some elitism, but I wonder if that is not a bad thing in this - where people love avatars with bad colouring and a outline of never removed white pixels in their gifs - or just those that lack any minimalism and use a mesh of stuff one can only pick up by a study of the avatar, something probably one shouldn't consider a good trait as one might view the avatar as a logo of sorts.
The best way I've seen to do this testing of avatars is by posting on deviantart.com with the avatars in question, and then see how many hits you get saying that it is good - the problem there is the large influx of pro anime users, not that anime in my opinion is bad, but anyone with a anime avatar will probably be given an boost. Anyway by doing this, and not begging people to look and consider the avantar, you will be given the reaction from someone who read the text, and like a normal users just noticed the avatar, and did not go in just to study the avantar in question causing the focus to shift like it does here in this thread.