Hacker News Reader

Comment by AnthonyMouse | Hacker News Reader

> In this lawsuit, Harvard is accused of directly discriminating against Asians, not just indirectly harming them by giving a boost to black or hispanic applicants.

But that's the exact same thing.

If you have a system where there are 1000 slots and the 1000 applicants with the highest scores get in, and you give Asian applicants +0, white applicants +5, Hispanic applicants +10 and black applicants +20, the outcome is entirely identical to giving Asian applicants -10, white applicants -5, Hispanic applicants +0 and black applicants +10. There is no meaningful difference, the people who consequently get in are exactly the same in both cases.

View on HN

Assuming scores are randomly distributed, your hypothetical gives different results if you give asians -5 than if you give black and hispanic applicants +5.

Harvard is accused of excluding more than half the asian applicants who otherwise would have made the cut. Harvard does not admit anywhere near enough disadvantaged minorities where giving those folks a boost would lead to that many asians not getting a spot.

Harvard (and other Ivies) repurposed a mechanism originally introduced by Yale to keep Jews out in the 1930s. If those Asian candidates had not been excluded using the arbitrary and capricious "personal" rating, it's mostly Caucasians who would have been displaced, not African-Americans or Hispanics.

The personal rating differs widely between the admissions committee and the the alumni volunteers (who give Asians roughly the same ratings as other groups). This happens for no other ethnic group. This is the smoking gun for racial discrimination.

Is there a source for this claim?

> Assuming scores are randomly distributed, your hypothetical gives different results if you give asians -5 than if you give black and hispanic applicants +5.

If there are only asian, black and hispanic applicants, it gives exactly the same results. When there are also white applicants, that change is equivalent to adding a -5 modifier for white applicants equal to the original -5 modifier for asian applicants. In neither case is the result a system in which race can only help you -- there are a finite number of slots.

It seems obviously wrong that if you gave +20 to white applicants it would not be a disadvantage to black, hispanic or asian applicants just because they all have the same equally-disadvantaged modifiers.

Clearly there is no system where you can admit a black or hispanic applicant, keep class size fixed, and not push out a marginal white or asian candidate. But that doesn’t mean there is no difference between “discriminating against asians” and “giving a boost to disadvantaged minorities.” In the former case, only marginal asians get pushed below the cutoff. In the latter case, mostly white candidates will get pushed out.

Especially in the latter case, the effect on asians is indirect and attenuted. For example, say Harvard would have been 60% white, 30% asian, and 10% underrepresented minority. You can double the representation of disadvantaged minorities by reducing the asian representation from 30% to about 27%. So basically the bottom 10% of asians get exluded.

What is instead being alleged here is that Harvard exluded more than half the Asians that otherwise would have been admitted.

> But that doesn’t mean there is no difference between “discriminating against asians” and “giving a boost to disadvantaged minorities.”

The only possible difference is adding white people (or some other separate group) to the "discriminating against" set. And the combination of that precedent and intersectional politics makes even those equivalent as soon as you start adding more non-asian subgroups to the set of disadvantaged minorities.

For example, the white population at Harvard is overly culturally homogeneous. It would legitimately be a significant benefit to their cultural diversity to get some more rednecks in there, using some suitably more specific definition with regard to income, hometown urbanization level, etc.

So if you want to discriminate against a specific group, easy -- just enumerate a bunch of underrepresented groups that don't include them and give them each a bonus. Any given applicant outside of the target victim group is likely to be in one of them.

> So basically the bottom 10% of asians get exluded.

That's just because you're using convenient numbers. If the merit-based numbers would have been 60% asian (e.g. because Asia has a couple billion people who would all like to go to Harvard), you do have to exclude half of them to get them down to 30%.

Reply on news.ycombinator.com