The Footprints of a ``Mastodon'': How a Decentralized Architecture Influences Online Social Relationships

Decentralized online social networks (DOSNs) have recently emerged as a viable solution to preserve the users' privacy and ensure higher users' control over the contents they publish. However, little is known about the backlashes that the decentralized organization and management of these platforms may have on the overlaid social network. This paper fills the gap. Specifically, we investigate how a decentralized architecture based on distributed servers impacts the structure of the users' neighborhood and their egonetworks. Our analysis relies on social data gathered from the decentralized micro-blogging platform Mastodon, the newest and fastest-growing decentralized alternative to Twitter. Our findings highlight that the social network supported by each server, namely instance, has a specific footprint in terms of degree distribution and clustered structure of the ego-networks of its members. Further, how users connect to people hosted in other instances is heavily bound by the server they are in. Moreover, users who tend to establish relationships in outer instances prefer to use a bunch of servers. Finally, we show that the ego-networks of the users are more clustered within the instance boundary, i.e. triangles are more likely to form among members of the same instance. All these findings suggest that the decentralization drives the social network to a structure that can be potentially very different from the usual one typical of centralized online social networks. Thus, the architecture of a DOSN is a factor developers and researchers should take into account when designing this kind of social platforms.