We study how information quality and network structure interact in a market where consumers choose whom to follow. Suppliers earn revenue from views and choose be- tween high- and low-quality content; consumers form links to both suppliers and peers, and prefer to share and act on high-quality information. We find that platform inter- ventions which raise quality when the network is held fixed can shrink or reverse once users adjust whom they follow. Verification tools, for example, make direct screening more informative; consumers respond by leaning less on peer sharing as a quality sig- nal, and if the peer network was doing most of the screening, quality can fall. Link choices under cost asymmetry also produce a core-periphery network structure: some consumers become hubs by following many suppliers, while others rely on these hubs rather than subscribing directly. This concentration has competing effects on qual- ity: routing attention through better-informed hubs sharpens screening, but the same cost asymmetry also leaves the high-cost type undersubscribed, lowering the volume of direct signals entering the peer layer, and the net effect depends on which force dominates. More broadly, information quality is highest at intermediate levels of social connectedness: added peer links first help high-quality stories separate from low-quality ones, but eventually make both qualities reach almost everyone. This parallels a result in fixed-network models, but here connectedness is shaped by users’ linking choices rather than fixed outside the model. These results imply that the network response to policy is a central consideration for platform design.

Download

This paper maps the collaboration network of English-language YouTube commentary creators and examines how a creator’s network position relates to their success. The dataset combines YouTube Data API records with 2,917 hand-verified collaborations among 1,391 creators, where a collaboration is a video featuring two or more creators each making a meaningful contribution. This definition is stricter than the subscription, comment, or hyperlink ties used in prior YouTube network studies. The network is sparse and highly dis- connected, with a core-periphery-like structure and slight homophily by race and gender. Degree and weighted-degree centrality are positively associated with view and subscriber counts; betweenness and eigenvector centrality show weaker associations. Channel age and total video count are also associated with success, with video count showing the stronger association. Many high-success creators sit on the network periphery without observed collaborations, suggesting that collaboration is one path to success among several. Because the analysis is observational and cross-sectional, the results are descriptive associations rather than causal effects.

Download‍ ‍SSRN‍ ‍ GitHub