the point is that holistic can be easily used as a cover for race based discrimination and that this is what is happening at harvard and elite universities. when you look at the quantitative disadvantage for asian applicants in terms of SAT/GPA, I think it's clear that homogeneity cannot explain this away. it's also a little absurd that they publicly place so much emphasis on diversity and "interesting" applicants when the actual student body of ivy league colleges is filled with bland wealthy strivers of all races.
the interpretation in your last paragraph is even worse in a way for faust and harvard: only children with the right parental financial resources and support would have been able to "Build an airplane, climb a famous mountain, invent something, be a sailor, race sled dogs" before the age of 18. it is also hard to see why an academic institution cares so much about such things.
this can be corrected for to some extent by weighting exam scores based on school/region/socioeconomic status. that would be fairer and more transparent than the current system used by elite universities in america.
the ability to develop an "interesting" application is of course strongly influenced by family background and finances...
in other words, use unclear and vague criteria to judge admissions that just happen to massively disadvantage asian americans. it's strange that Faust is not condemned for her absurd and racist suggestion that asian american applicants are on average much less "interesting" people.
perhaps michael wang fell short in terms of extracurricular achievements (though this is unlikely given information provided in the article). but this is unlikely to be true of all asian candidates as a group. serious questions need to be asked about the use of subjective and hard to quantify non academic criteria that discriminate against a particular group.
the usa is after all quite unique in using non academic achievements for admissions to elite universities...
perhaps i am missing something, but it looks like this just highlights two very common and easy to define types of error. how much practical value is there in this? Has this not been done before?