Discovered paper pair (Session 38). Detailed explanation not available.
An agent's position in a collaboration network determines their productivity. Higher network centrality increases output through strategic complementarities with connected agents. Activity in one domain (scientific) enhances productivity in another domain (technological), but this effect is asymmetric - the reverse direction shows no detectable influence.
view paper→Expert importance exhibits dual structure: relational importance (routing frequency, interaction centrality) diverges from intrinsic importance (causal influence via gradient attribution). Frequently-routed experts may act as structural hubs with limited functional impact, while sparsely-routed experts can be causally critical. Masking intrinsically-important experts causes disproportionate collapse in interaction topology compared to masking frequent peers, revealing hidden causal dependencies.
view paper→