Establish causal attribution of influencer sharing to observed citation differences

Establish whether the observed differences in citation counts for AI/ML papers shared by social media influencers (specifically @\_akhaliq and @arankomatsuzaki, as studied) are directly attributable to the influencers' sharing activities rather than coincidental correlations or unmeasured confounding, thereby providing conclusive causal evidence beyond observational analysis.

Background

While the authors find strong correlations and provide a causal inference analysis using negative outcome controls, they acknowledge the absence of randomized controlled trials (RCTs), which limits definitive causal claims. They explicitly state that without RCTs, causal attribution remains inconclusive.

This unresolved question emphasizes the need for stronger identification strategies to determine whether influencer activity directly causes increased citations, as opposed to reflecting selection effects, topic virality, or other confounders.