The Perfect Selfie, According to Science … and Art

Jeffrey Roberts
Vantage
Published in
4 min readNov 19, 2015

--

by David Schonauer

This news story comes in two parts.

Part one focuses on a computer science grad student, Andrej Karpathy of Stanford University, who decided to find out what makes the perfect selfie. Rather than consult Kim Kardashian, he used a Convolutional Neural Network (ConvNet, for short) to figure out the answer.

A ConvNet is a type of data-mining computer tool. Karpathy explains all in his blog post titled What a Deep Neural Network thinks about your #Selfie.

“ConvNets recognizes things, places and people in your personal photos, signs, people and lights in self-driving cars, crops, forests and traffic in aerial imagery, various anomalies in medical images and all kinds of other useful things,” writes Karpathy.

“Few of many examples of ConvNets being useful. From top left and clockwise: Classifying house numbers in Street View images, recognizing bad things in medical images, recognizing Chinese characters, traffic signs, and faces.”

In essence, Karpathy used his ConvNet to analyze millions of Instagram selfies, then analyzed the number of likes per follower, labeling the ones with the highest rates as good selfies, and giving those with the least number of proportional likes a negative label.

“Once it was clear his ConvNet parameters were working, Karpathy ran a batch of 50,000 previously unanalyzed selfies through the network,” notes the Huffington Post.

His findings: Being a woman makes for a better selfie. Being a woman with long hair that falls over her shoulders makes for a better selfie. Being a woman with long flowing hair who uses filters and composes with the rule of thirds makes for the perfect selfie.

So there you go.

Example images of good and bad selfies in the training data. These were given to the ConvNet as “teaching material”.

The second part of this story asks: Why do we care what makes the perfect selfie? A recent photo exhibition at the Ricco/Maresca gallery in New York sheds some light on the matter.

Called “Me,” the exhibition mixed together fine-art photography (including work by Elinor Carucci, Andre Kertesz, Francesca Woodman, Vivian Maier, Imogen Cunningham, and Berenice Abbott) with vernacular imagery. The show investigated what it is that distinguishes the selfie from the traditional self-portrait.

Andy Warhol

The New Yorker’s Adam Gopnick toured the exhibition and came away with some insights.

Two images by masters lend a clue. One, by Weegee, shows a police photographer posing a suspect who is seated in front of a white background, his mug-shot number pinned to his lapel.

“What is astonishing is how docile, agreeable, how mutual the transaction is — even in the municipal hoosegow our urge is to play along, to want to see what we looked like,” notes Gopnick. “The urge to be photographed — the desire to be photographed, and to look good in the photograph — can be stronger even than the urge to protest our own innocence or to get up out of the chair and head for the exit. A selfie is a mug shot we ask ourselves to sit for.”

Weegee

The other photo from the show — a 1978 Polaroid in which Andy Warhol is seen blowing his nose — reminded Gopnick that ease (via instant film or smartphone) is implicit in the true selfie. “[W]e, in effect, pay nothing for each additional image we take, and so we don’t have to be choosy about our remembered moments,” he writes.

Taking a selfie can be an act of narcissism, or, in certain hands, an act of imagination, Gopnick notes, citing the work of 12-year-old Kaia Miller, also featured in the show.

Kaia Miller

“The relentless appetite for selfies that, in a Kim Kardashian’s hands, is belittled can become, in the hands of someone so much younger and more ingenuous and earnest, a reminder that self-showing is not necessarily selfish,” Gopnick writes. “It can be a form of self-accountancy, of diary keeping, of journal making, even a sort of charming ritual of daily inventory.”

Elinor Carucci, from the exhibition “Me”

Originally published by AI-AP

David Schonauer is Editor of Pro Photo Daily and Motion Arts Pro. Follow him on Twitter. Jeffrey Roberts is Publisher of American Photography (AI-AP) the finest juried collection of photography in hardcover as well as Pro Photo Daily. Follow Jeffrey on Twitter. Follow Pro Photo Daily on Facebook.

Sign up for the free Pro Photo Daily Newsletter.

--

--