I don't think meritocracy implies cosmic fairness.
The way it's used by most hackers is to say that someone's work is judged only on its own merits, rather than the author's social status or how good they are at playing political games or anything else orthogonal to the work actually being valuable.
Judging people on their work is partly about fair play. But it's also justified by concern for the results. A system which rewards the people doing great work will produce much more of that than a system which rewards political connections.
I care quite a bit that the best cancer researchers should get hired, and the best ideas get the money -- their progress is likely to directly affect people close to me. And this is true but less obvious for almost every field... an economy where everyone was focused solely on zero-sum competitions would not grow very fast.
I don't quite know what cosmic fairness means. But perhaps cosmic unfairness involves people's talents not being identical. The goal of trying to get the fastest progress against cancer does not align with the goal of trying to remedy such unfairness.
To back up the other comment, recent studies find that students from worse secondary schools with worse grades end up outperforming students from more selective schools with better grades once in medical school.
So in order to get the best doctors you can't just be all "meritocratic" by looking only at the results, i.e. grades, but have to take other factors into account.
If our concern is with results, then we should be looking for the best predictor of results, and school type may well improve our predictions, as you say.
We should aim to admit to medical school the people who'll make the best doctors at age 40, not just the people with the best scores at age 18.
The problem carries through time, not just at the moment you're making the decision to invest. What about the better potential cancer researcher who was left behind because their parents were poor? What about the one that decided to go learn programming because they couldn't get the exclusive internship due to lacking connections?
The position that states "remedying inequity costs more than simply giving it all to the current best" is a shortsighted one.
The way it's used by most hackers is to say that someone's work is judged only on its own merits, rather than the author's social status or how good they are at playing political games or anything else orthogonal to the work actually being valuable.
At least, that's the ideal.