Is 'distinctiveness centrality' actually distinctive? A comment on Fronzetti Colladon and Naldi (2020).

Abstract

Distinctiveness centrality, which was proposed in 2020 to identify nodes that are connected to poorly-connected neighbors, is simply a minor variation on two existing centrality measures: beta centrality proposed in 1987, and gamma centrality proposed in 2011. In toy, empirical, and generated networks, I show that these three centrality measures yield identical node rankings under nearly all circumstances. Researchers seeking to identify nodes that are connected to poorly-connected others should not use distinctiveness centrality, and instead should use either beta or gamma centrality because they are more widely-known in the literature, are more flexible, and are computationally simpler. Additionally, researchers should be cautious when proposing new centrality measures, taking care to avoid duplicating measures that already exist.