- have people making the recommendations fallen in love with special models and methodologies
- were there dissenting results from various approaches and models?
- could the recommendations be overly influenced by salient analogies?
- have credible model and method alternatives be considered?
- do you know where the numbers came from?
If decision makers can manipulate data, models and parameters in real time they might raise additional questions back to the team. To validate (or discard) much more models, blazingly fast solvers and powerful hardware is indispensable.