Abstract
This paper collects the positions of four experts who participated in a panel on simulation-based policy support. The experts are well known for their contributions in this domain, predominantly for their use of agent-based approaches. The first section addresses the increasing requirements to integrate social modeling to support the evaluation of the socio-economic impact of policies, including questions of equity. Artificial societies are presented as enablers. The second section observes that the purpose of a simulation model is very much linked to its usefulness for supporting policy decisions. This implies requirements for learning agents and better representation of time and space. The third section focuses on the need to give agents more active behavior, to let them drive the action. While digital twin technology promises to help, the current state of the art seems insufficient. The section closes by looking at trust in simulation models, which will be needed for policy support.