3 ms·
> they're failures of judgement or knowledge by not relying on other facts or by relying too heavily on the tests. I think this is well put and I really apprec
by godelski 3y ago
> they're failures of judgement or knowledge by not relying on other facts or by relying too heavily on the tests.
I think this is well put and I really appreciate your response. I generally agree and I'm generally one that is quite critical of metrics. But I do not see this article as solely about a failure of test, but the interconnectedness and reliance on poor metrics rather than using them to update priors.
Particularly why I like the last sentence is I find this being one of the key points that distinguishes us humans from machines. Because we are able to do such things as "I see the rules, I understand the rules" and more importantly "I understand why the rules were made, and while this violates the rules it was clearly not something the rules were intended to prevent." Because rules __are__ made to be broken. Because rules are imperfect. Tests are imperfect. Literally everything is imperfect. We have a strong desire for order and perfection which helps us decrease the noise but I think we all could do a bit better at embracing the chaos a bit more.