And the final speaker in this session at the 2026 International Communication Association conference in Cape Town is Hannah Fecher, with a paper on the impact of the algorithmic environments of short-video platforms on political communication. Political actors have begun to adapt their content to these platform cultures (or to how they understand them) in order to reach constituents.
But content distribution is highly personalised and optimised to platform engagement, and some video characteristic are associated with higher vitality. Users also report a mismatch between viral tendencies and their own content preferences, however, especially also with respect to political content. This indicates that algorithmic amplification is often not related to individual user preferences but to broader trends, and also influenced by supply-side imbalances.
Users’ own reported preferences might also differ from their actual media engagement behaviours, especially in fast-paced spaces where decisions to engage or not engage with content are often made quickly with a simple swipe or scroll action. How does algorithmic visibility align with user preferences and user behaviours, then?
The focus here is on young users in Germany, and the project conducted focus groups with some 27 young users, to explore the political social media content types they were interested in; automated content analysis of some 32,000 actual posts by 1,900 political accounts on TikTok and Instagram; and an online survey of some 1,000 participants who were confronted with a selection of 1,100 videos form the larger collection, exploring engagement or skipping decisions.
These patterns do not always align: while users said they did not like negative political content, political attack videos were watched more often than expected, especially by male participants. Trendy content types like dance videos receive more views, but are not actively selected so often; they can be successful, but not if they occur too much.
Target group address features and affordances do not correlate with increased views; such features are not entirely ineffective, but other video features appear to be more decisive here. Obvious and artificial efforts to engage with younger audiences were not popular.
Algorithmic visibility, user expectation, and selection behaviours therefore differ substantially, the largest divergence is between evaluations on the one hand, and selection and algorithmic amplification on the other. Recommender systems reinforce the behavioural tendencies. This has consequences for political communication styles.
