[Paper Review] Critical Nodes In Directed Networks

Owen Sims, Robert P. Gilles|arXiv (Cornell University)|Jan 3, 2014

Game Theory and Applications31 references3 citations

TL;DR

This paper introduces a novel brokerage power measure to identify and quantify critical nodes—'middlemen'—in directed networks, where they control information flow between nodes. By leveraging the adjacency matrix and formalizing contestability, the method reveals that traditional centrality measures like betweenness and Bonacich fail to capture middlemen's true influence, as demonstrated in historical networks like Renaissance Florence’s marriage network and Krackhardt’s advice network.

ABSTRACT

Critical nodes or "middlemen" have an essential place in both social and economic networks when considering the flow of information and trade. This paper extends the concept of critical nodes to directed networks. We identify strong and weak middlemen. Node contestability is introduced as a form of competition in networks; a duality between uncontested intermediaries and middlemen is established. The brokerage power of middlemen is formally expressed and a general algorithm is constructed to measure the brokerage power of each node from the networks adjacency matrix. Augmentations of the brokerage power measure are discussed to encapsulate relevant centrality measures. We use these concepts to identify and measure middlemen in two empirical socio-economic networks, the elite marriage network of Renaissance Florence and Krackhardt's advice network.

Motivation & Objective

• To formalize the concept of middlemen in directed networks, where they act as brokers controlling information flow between nodes.
• To address the limitation of existing centrality measures—such as betweenness and Bonacich—in identifying nodes with strategic brokerage power.
• To introduce the concept of 'contestability' as a dual to middleman status, where uncontested nodes are middlemen due to lack of alternative paths.
• To develop a general algorithm based on the adjacency matrix to compute brokerage power for each node in any directed network.
• To empirically validate the model using two well-known socio-economic networks: the Florentine marriage network and Krackhardt’s advice network.

Proposed method

• Define a middleman as a node whose removal blocks at least one directed path between two other nodes, formalizing their role in controlling information flow.
• Introduce 'node contestability' as the existence of alternative walks between node pairs, establishing a duality: a node is a middleman if and only if it is uncontested.
• Propose a new brokerage power measure based on the adjacency matrix, capturing the extent to which a node mediates between others in directed paths.
• Construct an algorithm that computes brokerage power by analyzing reachability and path multiplicity from the adjacency matrix.
• Augment the brokerage measure to align with and subsume established centrality measures like betweenness and Bonacich centrality.
• Apply the algorithm to real-world networks, comparing results with standard centrality indices to assess performance and validity.

Experimental results

Research questions

• RQ1 How can the concept of middlemen be formally extended from undirected to directed networks?
• RQ2 What is the relationship between a node’s brokerage power and its contestability in a directed network?
• RQ3 Why do traditional centrality measures like betweenness and Bonacich fail to identify critical middlemen in directed networks?
• RQ4 To what extent does the proposed brokerage power measure outperform existing centrality measures in identifying influential nodes in empirical networks?
• RQ5 How does the structural position of middlemen affect information control and rent extraction in socio-economic networks?

Key findings

• The proposed brokerage power measure successfully ranks the Medici family as the most powerful middleman in the Florentine marriage network, aligning with historical analysis and outperforming eigenvector and betweenness centrality.
• In Krackhardt’s advice network, the method correctly identifies manager 21 as the strongest middleman (with highest brokerage power), while traditional measures misrank him below manager 18, who is not a true middleman.
• Node 15 in the advice network is identified as a strong middleman with high brokerage power, despite having lower betweenness and Bonacich centrality, demonstrating the method’s superior detection of strategic intermediation.
• The analysis reveals that 3/4 of the oligarch faction families in Florence are middlemen, compared to only 3/7 in the Medici faction, indicating greater structural fragmentation in the former.
• The study confirms that middlemen are not only critical for information flow but also possess rent-extracting power due to their uncontested position, validating their role as monopolistic brokers.
• The duality between middlemen and contestability is formally established: a node is a middleman if and only if it is uncontested, providing a theoretical foundation for network competition and control.