2021 Winner
Kein Filter für Rechts
Country/area: Germany
Organisation: CORRECTIV
Organisation size: Small
Publication date: 7 Oct 2020
Credit: Till Eckert, Alice Echtermann, Arne Steinberg, Celsa Diaz, Clemens Kommerell

Jury’s comments:
Excellent use of technology and data science to detect bias or propaganda in Instagram from the right wing. Even when it seemed like subtle advertisement, this talented team managed to detect how these extremists network and recruit. Creative storytelling is presented in chapters, each one explaining the findings that otherwise would not be easy to understand. The effort of analyzing thousands of accounts, more than 4,500, delivered in this case the expected impact: Many accounts and content were deleted from the platform, and good journalism once again enlighted society on how to be aware of this kind of dangerous manipulation.
Project description:
The investigative data research #KeinFilterFürRechts, enriched with conversations with insiders from the scene and ex-developers, gave the first comprehensive insight into the right-wing parallel world on Instagram, its most important figures, strategies and codes – and showed with impressive examples how little the company does against it.
Impact reached:
Instagram reacted immediately to the research, deleting a number of examples from the texts from the platform in the days following publication; German politicians took it as an opportunity to debate the topic of platform regulation.
Techniques/technologies used:
The team, consisting of several reporters, a data journalist and a researcher, observed the right-wing scene on Instagram for more than eight months. It developed several tools to collect and analyze large amounts of data on the platform (it had its approach peer-reviewed by two researchers).
The used tools reached from „Exponential Discriminative Snowball Sampling“ to one where we could see how strong connections between accounts on instagram are (based on follows) called „instaball“, to classical network analysis and text data mining. The process went through mulitple iterations including security and bias checks (the full process is explained here: https://correctiv.org/top-stories/2020/10/06/kein-filter-fuer-rechts-instagram-rechtsextremismus-daten-so-sind-wir-vorgegangen/#daten-daten-daten-so-sind-wir-vorgegangen)
Through the database, the team was able to map a network of Instagram accounts for the first time (4,500 accounts) and understand how right-wingers to far-right extremists communicate there and draw young people into their ideology. It turned out that Instagram’s algorithmic vulnerabilities are being exploited for this purpose.
What was the hardest part of this project?
One of the key challenges: The corporate philosophy of Facebook subsidiary Instagram prevents third parties from gaining access to the data stored on the platform. Only the online service itself and the users should be allowed to dispose of it.
For us, this meant that Instagram does not offer any way to collect data from accounts. We therefore had to get creative and find a solution ourselves in order to be able to answer our research questions. Which we did, because we believe that the public has a right to know about right-wing strategies on one the most-used social media platform these days.
What can others learn from this project?
Aside from the combinational usage of different tools in order to show a network on instagram, the way we presented to story deserves attention.
So, what was special about #KeinFilterFürRechts was not only the combination of several research paths, but also in particular the narrative format: The research was published in four parts over the course of a week, with the headlines of the individual stories and their publication date being teased on the website in advance. Interested readers were able to subscribe to e-mail reminders – which they also actively used. In addition, the team used #KeinFilterFürRechts (the research title, which sounds like a slogan but describes the research result very precisely), a highly frequented hashtag, under which thousands of people exchanged views on the topic in social media.
Project links: