Analysis of the early evolution of the Twitter (X) social network in Sub-Saharan Africa

Abstract In sub-Sahara African countries the proportion of the population using Twitter (now known as X) and other social media networks is growing. Understanding these networks allows us to understand changes in how people interact with each other, share information and carry out economic transactions. We hypothesise that the networks of sub-Sahara African Twitter networks have statistical properties consistent with being built on a principle of self-organisation: where decisions about who to connect to are less influenced by large commercial actors (as they might be in Europe, USA and other parts of the world) and are instead on the basis of local interactions between individuals. To test this hypothesis, we first collected data on the Twitter network of users in Tanzania. We found that the degree distribution followed a power-law with degree close to 2. We calculated path lengths, clustering, and assortativity of mixing for this network, as well as identifying the most influential users using eigenvector centrality. We then tested the degree to which these measurements were consistent with a variation of a friend-of-a-friend model of network attachments: where links are formed by individuals who join a network first identifying one individual at random (a friend) and attach to them and then choose n q people who the initial individual follows and attach to each of them with probability q . We found that for q = 1 and n q = 40 the model reproduces many aspects of the Tanzanian network, including the degree and clustering distribution. This model is not consistent with, for example, the USA or Japanese Twitter networks. Taken together, the model and its comparison to data from different real world network, supports the self-organisation hypothesis: a rule under which new members of the network connect to a random person and 40 people they follow reproduces many aspects of how the Tanzanian Twitter network has grown.