In a recent conference about data analysis, I highlighted
the importance of being well aware of one’s model’s assumptions. That all
models are wrong should not be a reason never to use them, but you must be
acutely aware of what their use implies.
I trust that you would agree with such an observation (OK,
some will not want to use imperfect model at all, but that is going too far,
and would have prevented any progress in human knowledge). But if you do, you
should be pretty worried. Because most developed countries (and probably even
more so less-developed ones) are now run on the basis of models whose
assumptions are wildly violated –to the point where the rules of such models
have made their way into their constitutions.
