It's interesting that in the 4 months since I've been here, the populations of animals and evil samurai faces seem to have deteriorated. They seem to be far less face-like and bird-like than before. Perhaps this is due to a larger group of human participants who have differing ideas and the pictures therefore are less able to converge...

In a certain way, MutatingPictures.com is currently still at around day 2 of the experiment. Why? Because the "human intelligence energy" or whatever you want to call it which the site received for computation on its launch day was *higher* than all the energy the site received in all days ever since *combined*. Which also means the original launch day experiment was never correctly repeated with any of the other sets.

I'm currently getting errors at Google Analytics for more precise number checking, but a typical day for the site these days results in 40 or so visits, whereas on launch day it was over 70,000 visits.

http://blogoscoped.com/files/mutatingpictures-traffic-2008-02.png

So there's not only the question of how we can undertake an experiment like this just once, but also: how do we make it interesting enough for repeat loyal visits? It was a bit of a battle against time with this experiment. Without fast interesting results, visitors won't come back... but without visitors coming back, no fast interesting results are possible.

Turn it into a captcha somehow?

I like Roger's idea. Instead of having your simple math problem to verify your not a bot, why not have one of your mutating faces pop up and ask the user how much does it look like a face. Would be kind of neat to see how quickly the faces start mutating again.

Colin
I am not sure what Roger mean by turning it into a CAPTCHA, but what you described doesn't sound like a good idea to me because: 1) asking "how much does it look like a face" is an opinion question, answer can't be definite, and 2) usability.

Yeah It wouldn't work cause any answer would be valid cause it was an opinion. Was just thinking out loud in that post. :p