What do algorithms see when they look at your social media profile? Are you a good provider of content? And what exactly is good? What does the algorithmic view on your social media profiles say about you?

Analyses of social media profiles are employed in an increasing number of real-world transactions. From border control to job applications, social media profiles are used to assess the threat-levels of a candidate and verify their suitability to enter a country or organisaton.

Many of these social media checks are done automatically by algorithmic entities. How and what do these algorithms assess exactly? And what do their assessments look like?

To find out, !Mediengruppe Bitnik used Ferretly, an automated online service used by Human Resources Professionals to evaluate candidates’ social media profiles. Ferretly uses algorithms trained on keywords and image recognition to sift through 7 years of social media history. The service evaluates candidates publically available social media posts on Twitter, Instagram and Facebook, analyzing original posts, reposts, replies, and likes. Additionally, Ferretly will assess any news items they can find.

With their service, Ferretly promises to reduce the risk of letting a person displaying toxic behaviour into your organisation or country. Toxic behaviour is defined in 11 risk categories, from hate speech, political extremism, drug-related content to explicit images and toxic language. The publicly available posts are also used to look at the sub-text of each post: the candidate’s attitude towards the event or situation they are posting about is rated as positive, negative or neutral. Each candidate is judged according to these per-defined flags and sentiment points and given a social media score which rates them as fit or unfit for entry.

!Mediengruppe Bitnik ran the social media profiles of the leaders of each of the 27 EU member states through the service. Bitnik then interpreted the results of the analysis and used this as the basis to devise a customized sweatshirt for each of the 27 profiles. 

Each shirt shows Bitnik’s interpretation and visualization of the Ferretly ratings. Besides their social media scores for different parameters, each of the shirts publicly displays a number of flagged posts which were rated by Ferretly as toxic and the reason for this rating. Depending on the report, the shirts contain more or less data. Usually more posts and data meant a worse social media rating by Ferretly.

Like the clothes we wear, our social media profiles have become the carrier of our identities. These online identities are used more and more by the gatekeepers of institutions, countries and organisations to verify that we are worthy of access.