In short

Gamma auction: 50 years, 50 prints in favor of Reporters Without Borders

An auction of photographs of the Gamma agency is held on Sunday, October 23 at 7PM at the City of Ar...

5th edition of the Photoreporter Festival of Saint-Brieuc

Created in 2012, the photojournalist Festival in the bay of Saint-Brieuc (Britanny) launches its fif...

Boutographies : Call for submissions, until November 13

As every year the Boutographies launch their call for submissions for its next edition, which will h...
Chelsea Girls, 1966 © Alan Aldridge
Chelsea Girls, 1966 © Alan Aldridge

Aditya Khosla, a computer science student at the Massachusetts Institute of Technology (MIT) has just published an article explaining how he developed an algorithm to predict the popularity of an image posted on a social network. No more breathlessly refreshing your homepage to watch the growing number of “likes.” Just let Khosla crunch the numbers and you’ll know in advance—on a scale of 1 to 10—about your chances of being a hit with your “friends.”

With nearly ten jargon-filled pages, the PhD candidate explains the highly scientific process used to understand what makes a vacation or cat photo popular—or not. He studied nearly 2.3 million photos shared on Flick in order to identify the characteristics of a “bankable” photo. He’s not messing around with the power of a selfie posted on Instagram.

Color plays an important role. Reds are more likely to win favor than blues and the  green. In fact, flashy colors are more likely to draw clicks.

Social context is also essential. Obviously, a more connected user will get more clicks than one with fewer “friends” or followers. Khosla lists all the items more likely to pull “likes.” You’ve been warned: spatulas and golf carts are no-gos. On the other hand, bikinis and revolvers are sure bets.

Without further ado, I decided to consult  the magic mirror myself, along with a few of the biggest contemporary photographers: “Khosla, Khosla, will I get the fairest likes on all the net?” Oddly, a kitten earned me the lowest scores. So I tried with one of Rineke Dijkstra’s girl in a bikini, then a William Klein photo of a boy with a revolver. I didn’t even break 5. What  with Andreas Gursky’s “99 Cents,” the most expensive photo in the world? A mere 4.674.

It was only by uploading my Facebook profile picture that I beat the average. It was a shot taken during a costume party where I was dressed like the characters Jean-Paul Goude invented for Kodak in the 1980s. In the end, it was a tribute to film photography that worked the best in the digital age.

What makes an image popular?
A. Khosla, A. Das Sarma and R. Hamid
International World Wide Web Conference (WWW), 2014