We don't need another decade building more complex models that exploit exascale computing, but one that:
1. Better understands & characterizes fundamental conceptual issues
2. Integrates multi-disciplinary knowledge & perspectives
1/
nature.com/articles/s4146…
1. Better understands & characterizes fundamental conceptual issues
2. Integrates multi-disciplinary knowledge & perspectives
1/
nature.com/articles/s4146…
Many presume that inadequacies of current models can be solved with more resolution, more detail, more computer.
But, fundamental questions on the inadequacies of models have note been addressed (eg model structure, initial conditions, nonlinear dynamics, etc)
2/
But, fundamental questions on the inadequacies of models have note been addressed (eg model structure, initial conditions, nonlinear dynamics, etc)
2/
"Climate economists [have] spent decades attempting to provide ever-better numerical estimates of a benefit-cost ratio... Even if the ECS isn’t strictly fat-tailed, the benefit-cost ratio [is] highly sensitive to ... parameters which suffer from deep uncertainty"
3/
3/
"Before ploughing billions into developing specialised computers & associated computer models, it would be wise to first develop a good theoretical understanding of what is necessary & sufficient to build models capable of such high-resolution predictions"
4/
4/
The article focuses mainly on climate modeling (eg regional impacts) & cost-benefit IAMs, but the points raised apply much more broadly.
Generally speaking, modelers (all too often) like generating & interpreting numbers, not understanding if those numbers are robust.
5/5
Generally speaking, modelers (all too often) like generating & interpreting numbers, not understanding if those numbers are robust.
5/5
• • •
Missing some Tweet in this thread? You can try to
force a refresh
