Comparing Centrality Measures

Centrality measures are tools to define 'importance' in a network, because being popular isn't the same as being influential. Think of it like a cocktail party: Degree Centrality is the person shaking everyone's hand (popularity), while Betweenness Centrality is the person standing in the doorway who decides who gets in (control). In our spy scenario, the 'Bridge' was more critical because they controlled the bottleneck between groups. Check this: In your next meeting, notice who simply talks the most versus who actually connects different teams.

To truly understand influence, you must look beyond who has the most friends. Network science offers four distinct lenses to view power:

Degree Centrality (Volume) measures immediate connections. It identifies the loud "hubs." In viral marketing, these are the influencers with millions of followers who can blast a message out quickly.

Closeness Centrality (Speed) measures the average distance to everyone else. These people are the most efficient broadcasters. In an emergency response team, the person with high closeness can get a message to the entire organization faster than anyone else.

Betweenness Centrality (Control) measures how often a node lies on the shortest path between two others. These are the brokers and gatekeepers. An executive assistant may not have a fancy title, but if they control access to the CEO's calendar, they have immense betweenness power.

Eigenvector Centrality (Prestige) accounts for the quality of connections. It’s not just how many people you know, but who you know. This is the logic behind Google's PageRank: a link from a major university website counts more than a link from a random blog.

Early sociological research focused on simple connectivity, but modern network science recognizes that 'centrality' is context-dependent. A measure that predicts infection risk (Degree) may fail to predict who controls information flow (Betweenness).

• Freeman (1979): Formalized the three distinct conceptual foundations of centrality: activity (degree), efficiency (closeness), and control (betweenness) [1].
• Borgatti (2005): Demonstrated that the 'correct' centrality measure depends entirely on the flow process (e.g., money transfer vs. gossip vs. disease) occurring on the network [2].
• Page et al. (1999): Introduced PageRank, a variant of eigenvector centrality, establishing that a node's importance is derived recursively from the importance of its neighbors [3].

Centrality measures are often calculated on static snapshots, missing how networks change over time. High centrality can also be a liability; in disease networks, 'hubs' are the first to get infected. Additionally, these metrics assume all edges are positive, often failing to account for antagonistic relationships or trust deficits.