Nothing in your post disagrees with what I said.
Two things, first you have to establish what discrimination means in a concrete way, and second you have to prove that Harvard is doing it. With regards to the first, the only statistically meaningful mechanism of measuring institutionalized discrimination is by looking for a difference between demographics of the institution's selection population, and the general population. With regards to point two, Harvard is obviously extremely discriminatory, but not to Asians.
Couple of more things:
Top Ten Things Overheard At The ANSI C Draft Committee Meetings: (10) Sorry, but that's too useful.