Spent 2.5 days in Eugene, Oregon learning about data, experiments and numerical models of the subduction zone megathrust. The workshop was organized as part of the Modelling Collaboratory for Subduction. The key take-away for me was a “long-term” subduction modeler is that the long-term stress state in the subduction zone is a key parameter that is imposed in earthquake rupture dynamics simulations. So, carefully including the compositional and rheologic layers at the plate boundary interface in long term models will allow us to “hand” the stress state to models of earthquake rupture. The other key take away is that we really need to build basics models of small-scale process like porous media flow that are appropriate for asking questions related to the megathrust environment.