OP touches upon a critical point of human reasoning: we have a bias towards simplicity. We favor the legible (as the article calls it), the tractable, the coun

OP touches upon a critical point of human reasoning: we have a bias towards simplicity. We favor the legible (as the article calls it), the tractable, the countable, the fungible. At certain levels of complexity, seeing reality as it is is just impossible. This complexity can be in terms of the subject itself, or the organizational/communication overhead required to tackle that subject. This bias towards simplicity is more often optimal than not. As he noted, the variance in performance becomes unbearably huge as you try to get more precise. The thing is that it's easy to pick out where simple models are wrong. The more complex the subject, the more edge cases there are, the fatter the tails are, etc. But it's harder to replace those models with new ones that increase precision while retaining sufficient efficiency. So just as simple models account for the cost of an employee without accounting for their value; he critiques those models for their precision without accounting for their efficiency. It's not as easy to predict/calculate value as it is to calculate cost. So while he's right in saying "these models are wrong", people won't be persuaded until they hear "these models are better". And "figure out the optimal solution in every individual context" is not a better model, not given the complexity of our work and the limited resources we have. Anyone taking the message here as "managers are idiots", are also oversimplifying things. They feel the moral high ground of being right, and it can be helpful to highlight where our models are wrong. But we also need to replace them with better ones, which is much harder than critiquing the best usable-but-imprecise ones we have.