Selfiecity is a newly launched project, co-ordinated by Lev Manovich and creatively directed by Moritz Stefaner alongside an ultra-talented team, investigating the style of 3200 'selfies' (photgraphed self-portraits) across five cities across the world. The project was previewed at Visualized conference but has now been let loose in the wild.
The project takes on the investigation of this contemporary phenomena by exploring a variety of attributes of the subjects, the poses and the expressions.
Up first, you can explore and view the images themselves via the ImagePlots panel, filtering by city and picking different cropping/positioning techniques. Alongside this we some demographic findings from the analysis of each and every photograph.
To gather data about the characteristics of age and gender for each person in each photograph (in supplement to 'rudimentary automatic face analysis with human judgment') the photos were inspected by Mechanical Turk workers. You can read more about the thorough data gathering and preparation process, including the process of refining an initial 120,000 photos down to the final 3,200 used in the study.
Next up, we have the Selfiexploratory, a separate interface where you can perform custom dives into the photos based on your own filters and parameters. For those of a certain age it feels like a digital take on classic 'Guess Who' board game...
Finally, we have the summary findings, providing insights such as the proportion of selfies out of the original Instagram image bank, the gender analysis by city, perceived smiliness (cheer up Moscow!) and the index of head-tiltiness (careful with those necks, Sao Paolo).
There are many things to praise in this project, which takes what would appear on the surface to be a relatively thin and superficial subject matter and successfully rummages and dissects its potential insights to the maximum.
I love the little green call-outs. The simple act of placing a revealable insight at different points throughout the site is a super device. The project is driven by curiosity ("Is it just me, or do Sao Paulo women actually tilt their heads more? Do New Yorkers or Berliners look older?"), not just opportunity ("We've got all these images, what shall we do with them"). It is also not just an explore-and-find project: it also incorporates explain-and-show. The creators have taken the responsibility to unearth and share their findings taking the experiment even to the realm of theory and reflection.