User Voting Behaviour in reward-based Social Networks

Cooperation is essential for addressing large-scale social dilemmas that require coordinated behavioral change. While traditional social media platforms offer valuable insights into cooperative behavior, their unstructured design and limited transparency hinder the systematic study of cooperation dynamics. In contrast, Web3 social platforms provide a novel opportunity to analyze online cooperation, as they embed monetary rewards that are explicitly tied to cooperative actions such as content co-curation. In this work, we study 2.5 years of user interactions on Steemit, a Web3 platform whose reward system inherently reflects a social dilemma. On Steemit both creators and voters of posts are rewarded proportionally to the post popularity, however users have a limited voting budget, encouraging strategic decisions about which content and authors to support through potentially different cooperation strategies. We investigate how individual voting behavior is shaped by both strategic and social factors under the platform's stable, transparent, and endogenous reward rules. We test three key hypotheses on whether cooperation is driven by reciprocity, familiarity with previous partners, or historical power of content creators in attracting high rewards. We find that reciprocity is the primary correlate of voting decisions, while content creator power plays a minimal role. This finding backs research in social psychology suggesting that psycho-social norms can exert a stronger influence on cooperative behavior than rational reward-maximizing strategies. Our study highlights the potential of Web3 platforms as empirical testbeds for cooperation research and offers practical implications for designing digital systems that promote sustained collective action.