Nicely put. This how I always think about economics: we've got plenty of numbers and models, but meaningfully large experiments are are to design without risking a lot of livelihoods. Science is brutal at scale, and I'm ever impressed with the techniques to make sense of the noise.