Humans are no longer the only actors in online social networks. There are a variety of computer-based and algorithmic actors broadly known as “bots.” Bots have become central to online phenomena such as social movements and open-source software, and are reshaping how we think of social actors in these situations. The increased dominance of bots in online networks, as well as the advancement of their underlying technology, necessitates understanding the differences in how bots engage with humans and with each other, and how these engagements change the overall dynamics in online social networks.
To unpack this puzzle, we begin from the premise that humans and bots attend to other social actors in fundamentally different ways. Human attention is a scarce resource, and therefore humans are limited to how many quality social relationships they can entertain. Because of this limited attention, there is generally an inverse relationship between the average strength of ties in an ego network to the number of ties. Although online social network platforms enable people to maintain a large number of contacts, central actors cannot sustain meaningful relationships with all contacts and end up using broadcasting strategies for interaction. The extended number of contacts enabled by online social networks does not necessarily translate to a greater number of strong ties – humans simply benefit from affiliations and cognitive efficiencies through their limited number of strong ties. Bots, on the other hand, do not have an inherent attention limitation. Bots do not realize the efficiency and long-term benefits of strong relational ties. Thus, in summary, we expect a tradeoff in range and tie-strength whereby humans will generally have greater tie-strength but less range, and bots will be the reverse.
This “blunt” characterization of the trade-off between range and tie-strength serves as a starting point. To begin unpacking the specifics of bots as social actors in online social networks, we report on a two-stage study of Reddit communities. The first stage involves exploring the distinctions between ego networks of humans and 500 bots, focusing on range and tie-strength for different sorts of bots and analogous humans. Range refers to the bandwidth of social actor attention. How many other social actors (alters) do they attend to at any point in time or over time? Tie-strength refers to the quality of each tie, in terms of its depth of interaction with alters.
The second stage of the study involves exploring the effects of bot involvement on community interaction through affiliation networks in a variety of subreddits. We characterize the impact of bots in terms of the overall cohesion of the community, theorizing that certain types of bots that act in ways that go beyond simple broadcasting to serve as a mechanism that reduces the modularity of the community by bridging structural holes. We further show evidence of community dynamics through the analysis of the frequency and density of local clusters through motif analysis.
