Modeling link recommendations as a network growth mechanism and their impact on social contagion

Link recommendation algorithms significantly shape online social networks, influencing both their structural evolution and critical processes such as the spread of information and behavior. This paper investigates how these algorithms affect simple and complex contagion processes by modeling link recommendations as a network growth mechanism. This allows for the direct comparison between algorithmic curation and organic growth processes like triadic closure and homophily. Using both heuristic and graph neural network-based recommenders, we find that complex contagions are highly sensitive to algorithmic influence. While moderate recommendation strength can amplify their spread, high strength suppresses it. This non-monotonic effect is the result of a structural self-reinforcement process, driven by top- k truncation: recommenders increase local clustering at the expense of global reach (average shortest path length) and strong truncation (small k) amplifies this effect. Our results underscore the nuanced interplay between network structure, recommendation strength, and contagion dynamics, highlighting the importance of incorporating social contagions into the design of link recommendation algorithms.