Résumé en anglais

The longitudinal study of advice networks among 240 judges at the Commercial Court of Paris permits the examination of learning as an interactive process. We argue that a spinning top model is a useful heuristic for intra-organizational learning in dynamic advice networks. This model proposes that a stabilized elite preserves accumulated knowledge in a community that overall experiences high turnover and systematic job rotation, and hence runs the danger of inadequately sharing knowledge among its members. We test the model by analyzing the structure and dynamics of advice networks among judges at the Commercial Court of Paris. This dynamic structure reflects the informal homophilous preferences among judges organized in a strong formal system, a high relational turnover in the selection of advisors, and the emergence of an elite of senior advisors that stabilizes the learning process - much like the behavior of a spinning top. This case study also identifies an endogenous process of increasing and then decreasing centralization of this network over time, raising questions about the maintenance of the stability of the pecking order and about the relationship between learning and seniority. Results illustrate the importance of dynamic over static network analysis and call for a renewed attention to formal structure in organizations.