This is the crux of the issue as I see it. The author sees achievement as a proxy for ability, and therefore achievement-meritocracy is the only acceptable system.
By contrast, MIT sees ability as only one factor among many that create achievement. If one takes the further axiom that ability is independent of protected categories like race and gender, then achievement-meritocracy does not recruit the most able students. Hence, affirmative action and reducing oppression.
Mathematically, author believes achievement=f(ability) so ability=f^{-1}(achievement) while MIT believes achievement=f(ability,identity, background) so ability=f^{-1}(achievement,identity,background)