> The tests that start as attempts to evaluate and quantify effectiveness become the goals themselves, rather than measures of the true goals. I think you need

> The tests that start as attempts to evaluate and quantify effectiveness become the goals themselves, rather than measures of the true goals. I think you need a second factor to make flawed metrics a big problem: increased competition. More competition strongly incentivizes all actors in a complex system to favor optimizing over the currently-measured metrics rather than the true goal. You see this playing out across all parts of society today: academics that chase impact score rather than groundbreaking research, bankers that chase short-term profits to the detriment of actual sound banking, and now we have a Navy that can't do basic seamanship. I think we really need to recognize as a society that if the thing you care about is inherently hard to measure (e.g. science, the overall stability of the financial sector, the ability to fight the next war while still at peacetime), there are no shortcuts, you just need to suck it up and keep trying to measure the real thing rather than only relying on proxies. Throwing more and more effort at optimizing over bad metrics will only lead to long term disaster.