In any network—social, biological, infrastructural—not all nodes are equal. Some people are more influential, some proteins more essential, some roads more critical. Identifying these key players is the central problem of network analysis. Degree, betweenness, closeness, and eigenvector centrality each capture a distinct notion of importance, and they often disagree about who matters most.
Degree Centrality
The simplest measure: a node's importance equals its number of connections (degree). The most popular person—with the most friends—has the highest degree centrality. Degree centrality captures local importance: how connected you are to your immediate neighbors. It's easy to compute but misses global network position entirely. A node with many connections to other peripheral nodes may rank high by degree but have limited actual influence over information flow across the whole network. Degree is a necessary but far from sufficient measure of true network importance.
Betweenness Centrality
Betweenness centrality counts how often a node lies on the shortest path between other pairs of nodes. Formally: C_B(v) = Σ [σ(s,t|v) / σ(s,t)], where σ(s,t) is the number of shortest paths from s to t and σ(s,t|v) is the number passing through v. High-betweenness nodes are brokers or bridges—they control information flow between communities. Remove them, and the network may fragment. In organizational networks, employees with high betweenness often wield disproportionate informal power despite modest formal seniority.
Closeness Centrality
Closeness centrality measures how quickly a node can reach all others—it's the reciprocal of average shortest path length to all other nodes: C_C(v) = (n-1) / Σ_u d(v,u). Nodes with high closeness can spread information or resources efficiently across the network. In supply chains, facilities with high closeness centrality minimize average delivery times. In disease networks, individuals with high closeness are efficient spreaders—a key consideration when designing targeted vaccination campaigns that aim to reduce transmission speed rather than just reach.
Eigenvector Centrality and PageRank
Eigenvector centrality captures the idea that connections to important nodes matter more than connections to unimportant ones. Your score equals the sum of your neighbors' scores—a circular definition resolved by finding the leading eigenvector of the adjacency matrix. Google's original PageRank algorithm is a variant: a web page's importance depends on how many important pages link to it, with a random teleportation factor added for mathematical stability. This eigenvector-based ranking made Google's search dramatically better than keyword-matching competitors when it launched.
Which Measure to Use?
The right centrality measure depends on what 'important' means for your problem. Identifying influential spreaders in epidemics? Betweenness and closeness matter. Finding the most connected hubs? Degree centrality. Ranking web pages for relevance? Eigenvector variants. Identifying critical infrastructure whose removal would fragment the network? Betweenness. In practice, researchers compute multiple measures and look for nodes that rank highly across all of them—these are the network's most robustly important players, important by any definition.
Conclusion
Centrality measures reveal the hidden architecture of influence in networks. Prominence is multidimensional: a node can be locally popular but globally peripheral, or modestly connected but strategically positioned between communities. By quantifying these distinctions, network scientists can identify key individuals for vaccination campaigns, critical infrastructure for protection, and influential nodes for targeted intervention in systems from the brain to the internet to global financial networks.