The Physics of Team Formation: Modeling the Catalysis of Collaboration at In-Person and Virtual Conferences*

The impact of interaction on the scientific process was brought to the fore as the COVID-19 pandemic forced gatherings to shift to a virtual setting. In this talk, we present a new nonlinear dynamical model for the origin of scientific collaborations at conferences, inspired by the physics of catalytic processes.  Our model takes as input the pattern of interactions among participants and predicts the probability that any given team will form. We test the model with a novel dataset tracking multi-year “Scialog” scientific conferences, including room-level participation data from 12 in-person and six virtual meetings, each with about 50 participants. Our model agrees well with data and outperforms seven other candidate models.

 

This talk will also cover new empirical evidence for the causal role of interaction---including prescribed interaction---in subsequent team formation, and will touch on the question of whether virtual conferences can effectively replicate the in-person experience. Our work illuminates a key role conferences play in steering the direction of scientific endeavors, and could help organizers optimize future conferences, regardless of their format.