Hugely important point, and something that we do NOT do today very often — and why what you are…
Henry Kim

One thing that I should like to add is that, while the kind of rejection of theorizing/model building in favor of blind worship of practical data as embodied by the Wired article is disturbing, it also does point to where the useful balance can be struck. The trouble with “theory” is that humans are too quick to form theories and too slow to abandon them, while, for the machines, somewhat the opposite.