CAGE Work in Progress

Title: Brainrot? The Effect of Short-Form Content on Long-Form Attention

Abstract: Has the rise of short-form content worsened our attention spans? I assess this question by studying slice-by-slice attention to news videos on YouTube, the leading global source of video content. Leveraging the staggered rollout of YouTube Shorts across markets and channels, I find that the introduction of short-form content significantly worsens user attention to their usual videos. Skipping to any given section of a video increases by approximately 30%, with across video comparisons showing that longer videos are worst affected. The effect is fast-acting and permanent. On the supply side I find no evidence that channels modify the content of their videos, but instead increasingly produce their own short-form content. Disaggregating this skipping within videos, I find a uniform increase to all slices except for the final 5% of a video. I rationalise these results with a model in which videos are not discrete units but a collection of interdependent slices. Skipping allows users to trade off context for additional video content, with shorts exposure lowering the value of context and leading to content-seeking behaviour. Structural estimation reveals that viewers value context 15% less relative to content in response. Mechanism tests suggest these results are driven by a worsening attention span, rather than substitution or user composition. Overall, short-form content has large impacts on user behaviour.