What can we learn from selfies? As it turns out, quite a lot. What started as a vanity project, another excuse for a hashtag, or perhaps just a natural progression of the social sharing trajectory is now much more - it's a rich mine of data, perhaps best illustrated by the Selfiecity project.
Kicking off at the tail end of 2013 (roughly when the very term 'selfie' was crowned Word of the Year), Lev Manovich, Daniel Goddemeye, and Moritz Stefaner's Selfiecity aimed to break down the facts, figures, and fictions of the infamous #selfie. The Selfiecity crew has some serious pedigree in the digital space, too, with Manovich the brains behind Phototrails and a professor of computer science at The Graduate Center, CUNY, Goddemeyer a researcher and strategic interaction designer, and Stefaner a go-to guru of data visualization.
For a week, the team pooled around 656,000 photos from the Internet's prime selfie repository, Instagram. After collating photos from five cities (Bangkok, Berlin, Moscow, New York, and São Paulo), the team then whittled the collection down to a total of 640 images, each of which were then examined closely for exhaustive analysis of context and cultural data. The findings illustrated a variety of selfie habits (which typically differed between countries), including the age and gender of the photographer, the prevalence and degree of the 'awkward lean' (also known as a head tilt), and the emotions captured.
Interestingly enough, despite some 79 million photos bearing the hallmark hashtag, selfies only account for around three to five percent of total Instagram photos. There is, of course, some wiggle room: according to Wired, other hashtags for similarly inclined photos include #selfies, boasting around 7 million photos; #selfienation with 1 million photos; and #selfiesfordays with approximately 400,000 photos. There's also an unknown number of selfies that don't play host to any of the above hashtags, which may skew the data somewhat. Nevertheless, the low number came as a shock to the team behind Selfiecity. "I'm still surprised the number is so low," said Manovich.
Though the data reveals no great shocks, plenty of broadly acknowledged assumptions now seem to hold water, with the findings supporting certain theories. "The idea was to confront the generalizations about selfies, which are not based on data, with actual data," confirmed Manovich. "We wanted to look at what the actual patterns are."
For one: selfies are a young person's game, with the estimated median age clocking in at 23.7. Women generally take more selfies than men (in Moscow, it's an 80:20 ratio), though at the age of 40, there's a bit of a switcheroo, with men becoming more inclined to self-photograph. Similarly, women were found to be more likely culprits when it came to the awkward lean, tilting their heads a whopping 150 percent more than the men surveyed - that is 12.3 degrees to men's 8.2 degrees. In São Paulo, women are the most emphatic head tilters, clipping the scales at 16.9 degrees. Conversely, ladies in New York opt for a more subtle lean, reaching just 11 degrees.
Mood analysis also formed a key component of Selfiecity's data mining, with Moscow residents turning in the fewest smiles in their photos. The residents of Bangkok and São Paulo, however, compete for the top spot, or at least the most outwardly content photo faces. Selfie enthusiasts in New York and Berlin vied for the middle spots. Of the surveyed cities, New York appeared to be the most ethnically diverse, living up to its reputation as a hub of multiculturalism. However, as Manovich pointed out in an email to The Guardian, it proves difficult to identify a tourist from a resident. "We can use some of Instagram data to guess this - but there is no way to know this exactly in many cases," he wrote.
With selfies showing no signs of losing traction, we're sure there's many more ways in which the information can be gleaned and interpreted, leaving us to learn new facts along the way. Perhaps the duckface photos will even find their way into big data, providing a fresh perspective from which to analysis patterns in human behavior.
