Physics of social interaction at virtual and in-person conferences: a model for the formation of scientific collaborations

The COVID-19 pandemic has sparked a growing debate about the value of in-person compared to virtual interactions. Here, we show that properly designed virtual meetings generate novel collaborations. We present a nonlinear dynamical model for the origin of scientific collaborations at conferences, inspired by the physics of catalytic processes. We tested the model using data we constructed as part of a longitudinal dataset of “Scialog” conferences, including room-level participation data from four in-person and six virtual meetings, each with about 50 participants. We show that interaction in assigned groups was a better predictor of who ultimately formed teams at virtual compared to in-person scientific conferences. We attribute this to informal interaction playing a greater role in team formation during in-person rather than virtual meetings. This observation is supported by anaysis which shows that in-person conferences strengthen network connectedness more than virtual conferences. This suggests that virtual conferences are better at engineering team formation, while in-person conferences are superior for strengthening participants’ awareness of each other.