Hierarchy and Collective Intelligence

Abstract

It is often observed that superiors within an organization can lead subordinates because the former are smarter or have more organization-related knowledge. In this paper, we argue that even without such superior talent or prior knowledge, superiors can outperform and lead subordinates. Combining empirical data for a hierarchical network, we develop computational models of learning to validate our argument. Our results show that transient learning performance is systematically higher for individuals of higher rank in the hierarchy than for those of lower rank. This remarkable order in learning dynamics arises from the structural regularity inherent in typical hierarchical structures, where superiors are reachable to other organizational members by a fewer steps in the formal chain of connections than their subordinates and where rank determines the scope of an actor's access to various subunits. Our numerical analysis shows that different ideas and knowledge are primarily exchanged through superiors who have more exposure to diverse sources of organizational knowledge across different subunits. The implication of our findings is that superiors do not have to be smarter or more knowledgeable about organizational tasks. They can have a positive influence on organizational learning if they know how to leverage their positional advantage, which is embedded in the hierarchical network.

Keywords: Learning, Hierarchy, Networks, Complexity, Authority