I’m a mentor for Nina Jones for BBC Material World’s contest, [“So You Want to Be a Scientist”][BBC]. We set up a survey on the project [Facebook page][FB]. The survey is finished, the data is almost cleaned and now Nina is getting ready to code the profile pictures we’ve captured. Perhaps the name of the contest should be altered, however, as I’ve been having an awful lot of fun with the profile pictures: “So you want to be a graphic designer”, perhaps? Below is a photo of all the thumbnails ordered by saturation. A big version is here
Now we said that we didn’t want to uniquely identify people, but we would show profile pics in reports and presentation. So I originally had ordered these by Facebook ID, but then figured that is in some way a form of identifying information (outside the picture itself). Being unsatisfied with this, I came up with a simple saturation ordering and voila.
I did most of this work in the [Python Imaging Library][pil] (PIL), which unfortunately, doesn’t play nice with Mac OSX. It involved first getting an ‘average color’ by shrinking the image down to one pixel, and then sorting the image by the sum of the RGB color values. Then, I simply stacked all of them (all 1708 images that is) in rows of width 50 to get as close as possible to the dimensions of A paper.
Finally, although it is called “The Many Faces of Facebook”, I will accept that those faces look awfully white. Of course, these photos are drawn primarily from people in the United Kingdom (who are primarily white) and primarily from listeners of Material World on BBC Radio 4. This is in no way supposed to be a unbiased sample of Facebook, so please take it with a grain of salt and enjoy.
Special thanks to Terry Handcock, who came up with [the algorithm for finding the average color][py].