> if you judge everyone by (objective) standard tests, Goodhart's Law takes effect, and we breed a generation of highly optimized test takers.
As opposed to a generation of people highly optimized for a fuzzier metric that nevertheless exists. I know a bunch of graduates from elite universities who all just happened to be really into debate/mock trials/mock UN, competed at the national/international level at obscure sports, founded a school newspaper (we had five new school newspapers every year because everyone wanted to be a founder, few of them last once the founder graduates), "founded" a "business", etc in high school[1]. They're all in consulting/banking/law/ML now. Mildly obfuscating the metric doesn't actually make Goodhart's law go away, it just means you have to be in the right crowd to game the metric.
[1] I, on the other hand, actually liked maths and became one of your "highly optimized test takers". Funnily enough, all the elite American schools said no, while Oxford said yes. I am now a PhD student (in maths) at one of the elite American institutions that rejected me as a high school student.[2] If you have a scarce resource, and a system that distributes it, people are going to game the system. The trick is to have better tests so you can't study to the test without actually, you know, learning.
Take a look at the SAT (subject) math section. That's the test that should be distinguishing between the student that goes to Harvard and the student that "merely" goes to Georgia Tech. Obviously it can do no such thing, it's way too easy. This is by design. The American system does not want to distinguish between the top 3% and the 0.5%, because then Harvard would have to take the poor genius over the solidly talented but unexceptional rich legacy kid whose dad is worth billions and will donate tens of millions.
Now take a look at the MAT, the maths test that tells you if you'll be going to Oxford or Imperial. Or the STEP, the analogous test for Cambridge.
[2] Disclaimer: I may be a little bitter to this day. In my defence, I teach the wannabe consultants now, and while some of them are legitimately very competent, boy does the system have false positives.