I make my living predicated on the assumption that digital optimizers over loss functions have practical value, but seeing a pretty terrifying meta-pattern wher

I make my living predicated on the assumption that digital optimizers over loss functions have practical value, but seeing a pretty terrifying meta-pattern where defining and then subsequently obsessively optimizing and inevitably gaming metrics will burn away basically all of the useful parts of most human institutions, as we see in nearly every kind of business as noted here, as well as government endeavors like education, I'm pretty close to genuinely believing that quantifying success and trying to optimize over a metric is something human institution should literally never do, which is a weird place to be