YouTube favored far-right-wing videos in the run-up to the Finnish presidential election in 2024, claimed journalist Clothilde Goujard in Politico magazine on the 26th of March 2024. Goujard bases her claim on a well-known Finnish factchecker’s report. We checked the claim and found it uncheckable. According to experts, the data collecting period was not long enough to provide reliable results.
In Politico’s article “Europe wields new tech law to protect EU election”, journalist Clothilde Goujard claims that YouTube favored far-right-wing videos in the run-up to the Finnish presidential election in 2024. The article discusses the European Union’s new Digital Services Act (DSA).
According to Goujard, the claim is based on a study conducted by a Finnish fact checking platform Faktabaari and software company Checkfirst. The study tracked the bias of YouTube’s recommendation system in Finland’s 2024 presidential election. Faktabaari is a Finnish fact-checking and digital information literacy service and CheckFirst is a Finnish software and methodologies company.
Faktabaari’s and Checkfirst’s study was conducted due to concerns that YouTube’s algorithm may lead people to misleading or otherwise harmful content. YouTube’s recommendation system is built to keep viewers on the platform as long as possible by recommending videos one likes to watch. The Crossover report was published in March 2024 and is part of the Digivaalivahti, a digital election project.
Data Collection Using Predefined Terms
According to the study, data was collected from different parts of Finland. The data collecting computers performed searches using 77 predefined search terms and stored ten “watch next” recommendations for each individual video in the search recommendations. Data was collected from all over Finland to limit the regional recommendations in the results.
The data was collected by computers located in different parts of Finland.
The study concludes that YouTube’s recommendations were limited and gave visibility to the policies of the Finns party in particular. The Finns is a populist party that can been described as “illiberal radical right” (linkki: https://theloop.ecpr.eu/the-new-leader-of-the-finns-party-reinforces-an-illiberal-turn/), and part of European Conservatives and Reformists (ECR). It is Finland’s second largest party. Furthermore, Faktabaari’s study suggests that YouTube’s recommendation system can have unintended problematic effects if it does not provide users with a diverse range of content.
According to Pipsa Havula from Faktabaari, the study merely looked for the names of far-right politicians in the titles and descriptions of the videos. The content of the videos was not analyzed further. In this study, the term “far-right politician” specifically referred to members of the Finns Party. Essentially, any content that included the name of a Finns Party member was classified as far-right in the study.
Google: Misleading Conclusions
Google Finland’s Head of Public Policy Heidi Jern reacted to Faktabaari’s research by a letter to the editor that was published in Finland’s biggest newspaper Helsingin Sanomat on 14th April 2024. She criticized the study’s methodology as flawed and its conclusions as misleading.
We contacted Jern to ask her what she meant specifically. She responded that the YouTube algorithm considers many factors, such as likes and dislikes on videos, watch times and online habits. According to her, researchers at Harvard and the University of Pennsylvania have found that echo chambers are not a result of YouTube’s algorithms.
Faktabaari’s researcher Aleksi Knuutila responded in Helsingin Sanomat on 19 April. The fact checking platform admits that they did not analyze the content of the videos but looked at the prevalence of politicians’ names in the videos. However, according to Faktabaari, it was clear that YouTube’s recommendation system highlighted some far-right-wing politicians.
Expert: Data collection period too short
We consulted Dorota Glowacka, an associate professor in the department of computer science at University of Helsinki. She analyzed the report done by Faktabaari and CheckFirst and gave us her insight about the methodology used in the research.
She pointed out that the devices used to collect data were placed in various people’s homes, but the report didn’t explain how these homes were chosen or if the participants were volunteers. This could lead to bias based on their socioeconomic or geographic backgrounds. Dr. Glowacka also wondered whether other people living in the same house or using the same IP addresses could affect the recommendations.
The timeline of the research interested Dr. Glowacka as well. The data was collected over a three-week period, and Glowacka questioned if it was enough to provide reliable information. To notice the variations in recommendation patterns or find out long-term trends, the data should’ve been collected over a longer period.
She concluded that the methodology was broadly correct but would require much more time and effort to produce reliable results. Because YouTube is a part of the larger Google ecosystem, she thinks that its recommendation system could also be affected by bigger trends and world news. This is why it is difficult to identify why the recommendations might be biased.
Previous research
The research done by Faktabaari and CheckFirst is not the first one to investigate YouTube’s recommendation system. Researchers from Princeton University and UC Davis pointed out that following the recommendations by YouTube could result in radical videos even though they aren’t shown on the frontpage of YouTube. Earlier research published by PNAS Nexus, an open access journal, claimed that the recommendation system leans more towards left-wing content. These differences were talked about in research by Hosseinmardi et. al. and they discovered that the biggest reason for the various results was the difference in methodologies.
YouTube has ensured that they aim at decreasing the visibility of borderline content. This includes videos that spread misinformation. In addition, the new Digital Service Act (DSA) by European Union requires big platforms – like YouTube – to address risks that could destabilize, for example, media freedom, freedom of speech or general safety. The regulation took effect on all platforms on the 17th of February 2024, so it wasn’t in force during Finland’s presidency elections in 2024.
Conclusion
In our research we find that the claim “YouTube favored far-right videos during the 2024 Finnish presidential election” is uncheckable. The data was not collected for long enough according to the expert, and the content of the videos was not analyzed. It is therefore not possible to say on what basis the videos were recommended. In the light of these facts, it’s not possible to make direct conclusions that YouTube has recommended right-wing content in the run-up to the Finnish presidential elections.
RESEARCH | ARTICLE © Jasper Kalpio, Emma Kärki, Veikka Lokka, Veeti Välimäki, Haaga-Helia University of Applied Sciences, Finland
Leave your comments, thoughts, and suggestions in the box below. Take note: your response is moderated.
