The biggest point in this stands out as follows: The gatekeepers of "Merit" rely on <Hueristics> for selection. They do this precisely because they do not know actual merit. Cue: Irony. Applications to top-tier selective Universities, for example, are read for ~20 minutes. The reason they do not know actual merit, is a fundamental issue: Bounded Rationality.

The costs (in time, money, and brain damage) associated with "factual" merit testing would be rediculous/uneconomic. We can verify this somewhat orthogonally, by looking at a couple different similar selection processees. eg. Job interviews. Same problem, same solution. 20 min resume screens. Next, we can compare a resource-unconstrained organization: How does it recruit top 1-5%'er? (special forces, test-pilots, and astronauts, etc): long, exhaustive, actual-failure-mode testing. No 20 minute skims.

If every job or educational credential was allocated in this manner (and you'd have to assume access to the test and test prep), it would be a more interesting test of "Merit". As a second best, it would be interesting if schools actually failed out their weakest students (ie, proving that only the best had a credential). But that is not how the world works. Merit is, rather, a politically correct (and practically expedient) heuristic for preferential treatment. And that is the main point of this post, I believe.

The fair debate, of course, is on the resolution and applicability of filters under various contexts. And if they are inefficient, are they remdiably so?. &tc.