Why spreading knowledge across members can beat concentrating it
Distribution is the other half of the picture: which member holds which piece of knowledge. In self-directed teams, leadership perceptions that were distributed and coordinated across members were associated with team mental models and performance, in contrast to arrangements where leadership was not spread in a coordinated way [1]. The word coordinated matters here. Distribution by itself is just fragmentation; what appears to help is distribution that is organized so the pieces fit together.
Work on how groups use distributed information points in the same direction. When important decision-relevant information is spread across group members, shared task representations appear to help the group bring that information together rather than leaving it stranded with whoever happened to hold it [3]. That study's authors caution that their findings seem mainly applicable to groups dealing with distributed important decision information, so the result is scoped to that kind of decision setting rather than to all teams [3].
How similarity and distribution work together
The two features are not rivals. Similarity supplies the common frame; distribution supplies the different pieces to put in that frame. A team whose members share a task picture but all know the same things has little to combine, while a team with diverse knowledge but no shared frame may struggle to recognize which member's information is relevant. The self-directed team findings are consistent with this pairing: distributed-coordinated leadership was linked to both mental model similarity and performance, suggesting the arrangement of knowledge across members and the overlap of their mental pictures travel together [1].
Treating team knowledge as having a dimensional structure and a network representation offers a way to describe this pairing concretely. Rather than reducing a team to a single average of shared knowledge, this approach separates how much knowledge is shared from how similarly it is held, and connects those properties to performance and decision-making [2]. That separation is what makes it possible to ask whether similarity and distribution each contribute, rather than collapsing them into one number.
What these studies do not yet settle
The evidence supports a directional answer, not a precise recipe. The findings on distributed-coordinated leadership come from self-directed teams working on a task the authors describe as complex, so the pattern may depend on task complexity rather than holding universally [1]. The distributed-information findings are explicitly limited to groups handling distributed important decision information, such as top management teams, by the authors' own note [3]. And the work on knowledge structure is framed around measuring and representing team knowledge rather than testing a causal mechanism [2]. So the honest summary is that similarity and distribution appear jointly relevant to coordination and performance, with the exact balance likely depending on the task and the team.
Sources used in this answer
The impact of shared leadership on teamwork mental models and performance in self-directed teams
In self-directed teams, distributed-coordinated leadership was associated with team mental model similarity and team performance, in the context of a complex team task.
Team knowledge: Dimensional structure and network representation
Team knowledge can be described through dimensional structure and network representation, separating shared task knowledge from knowledge similarity and relating these properties to team performance and decision-making.
The use of distributed information in decision making groups: The role of shared task representations
Shared task representations appear to help groups use distributed information in decision making, with the authors noting the findings mainly apply to groups dealing with distributed important decision information.