And the next speaker in this session at the ECREA 2026 conference in Brno is Felix Gaisbauer, whose focus is on the supply of and demand for political content on TikTok. This is often tested here using sockpuppet studies that test what kind of content the TikTok algorithm serves users; here, sockpuppet accounts that displayed a right-leaning accounts have at times been found to receive more recommendations for politically aligned content than right-leaning accounts. This could also simply result if there is more right- than left-leaning account on the platform, though.
This ‘algorithmic supply’ thus intervenes between content supply and user demand, but in itself is also affected by economic imperatives, since advertisers are also paying TikTok to steer user attention towards their content. This means that TikTok serves as a multisided market – and TikTok’s specific setup means that its content market affords virality out of nowhere in order to sustain broad supply and provide an infinite scroll that is designed to maximise the time users spend on the platform.
To study this requires several parts, then: it needs an assessment of political content supply, via video scraping using specific keywords or producer lists; it needs information on what political content is actually served to users, which can be done via data donations; and in combination (and following LLM-based classification of the content) this can then be used to determine how political video supply and demand line up against each other.
In the context of the German election of 2025, videos about the far-right AfD dominated the political video supply; users were also disproportionately confronted with such videos, but AfD opponents mostly received videos which were critical of then AfD. Whether such videos were positive or negative, though, did not affect whether users watched such videos all the way through, though they predominantly liked videos whose perspectives they agreed with.
Data donations can this be utilised to disentangle user demand and algorithmic supply; it remains unclear whether user demand drives algorithmic supply or vice versa, however.
