What comes after Chapter 15?

Abstract:

"All models are wrong but some are useful"

In this panel discussion I will discuss thoughts on what it takes to make space environment models more useful than they are today. In what I have sometimes told my students in thinking about what comes after "Chapter 15" in textbooks, developing space environment models to be more useful requires going outside of the prescribed methods often taught in graduate school classes. It's clear today that when you dig deep into current state of the art operational models they were developed at a time where making approximations were a necessity because of the limit of computation at that time. Those decisions made back then are still with us today. However when we look at current and future computational capabilities (high performance computing, GPUs/TPUs, machine learning techniques, etc) models should now be built, starting from the ground up, incorporating all known relavent physics, from the micro to the macro. Tools like machine learning now allow us to bridge the problem of scales and the dawn of exascale computing makes truly first principled modeling more realistic. Of course this means that research groups can no longer develop models that produce results that meet the "minimum publishable unit", rather, long term and high risk planning should take precedence so that the developed model becomes useful to the end user.