[downut]

Now fit legacies into this theory.

The wikipedia page on Legacy Preferences is illuminating. Note the Larry Summers quote:

Former Harvard University president Lawrence Summers has stated, "Legacy admissions are integral to the kind of community that any private educational institution is."

[credit_guy]

Just about a month ago I realized for the first time that legacy admissions result, in many cases, in better candidates rather than worse. Not always. But here's an example (from real life, without names obviously): highly qualified student, with lots of national level achievements, Cornell legacy. Applied in the early admission period to Cornell, got in. But the student had a reasonably high chance for Princeton or Yale, let's say. However, the legacy system incentivized him to apply to Cornell, even if his level was slightly higher. Why? Because if he didn't apply in the early period to Cornell, hoping for Princeton or Yale, and didn't get in, then Cornell would not have given him any preference in regular admissions. So he had to choose between nearly 100% admission at Cornell in the early round, vs 10% chance at Princeton, and then a non-negligible chance to not get into Cornell in regular.

My point: legacies are not always dumber than non-legacies. Sometimes they are stronger, and the legacy system incentivizes them to stick to the school where they are legacy.

[only-one1701]

Which kind of student do you think would be more likely to realize that a single data point is of no consequence: legacy or non-legacy?

[credit_guy]

Nice roast there, but was it really necessary?

[only-one1701]

I think it’s a pithy counter to the point you were trying to make.

[credit_guy]

Sure. It also sounds like you liked the pithy answer opportunity more than you than you care about the substance of what I was saying. Which is a shame, the dialogue on Hacker News is generally better than that.

[only-one1701]

No, I believe the pithy answer also conveys a sufficient rebuttal to the "substance" of what you were saying. Or to put it more explicitly: the scenario you lay out is rare enough to be of no consequence compared to the much more frequent instances of unremarkable legacies getting a massive handicap.

[onetimeusename]

Let's say two people are applying to Harvard and it's the year 2019 (I think they stopped legacy admissions recently). They both have a 1550 and 4.3 GPA. Both went to good high schools. Both helped underprivileged youths learn to code. However one of them has two alumni parents who are both well known, Pulitzer Prize winning journalists in DC that helped expose the corruption of the much hated X politician during the Y scandal, they are White House Correspondents and they are regularly featured on the news. The other student has parents who are not alumni. Harvard has to pick between these two students. Which one do you think Harvard picks?

Note that you cannot argue the legacy student actually has a lower SAT score because Harvard admitted legacy students had higher than avg. SAT scores and because the study controlled for SAT score.

Believe it or not, this is the kind of profile a lot of legacy admits would have.

[downut]

"Which one do you think Harvard picks?"

Do you not understand what the point of legacy preference admissions is? I will supply it here: Legacies take the place of the higher performing non-legacy candidate, not the equivalent one. Is this difficult to understand? Why?

[onetimeusename]

I don't think that's the point of legacy admissions. I think it's purpose is exactly what the grandparent said which is to cultivate a network of people.

The problem with saying legacy preference is to take the place of higher performing non-legacy candidates is that legacy admits generally perform above average at least at Harvard although it's probably true elsewhere. See here[1]: The average SAT score among legacy students was 1543, while it was 1515 for non-legacy students.

So it could still be for replacing higher performing non-legacies meaning Harvard targeted and rejected a bunch of people with even higher SAT scores than the legacies but I don't find that very convincing.

[1] (https://features.thecrimson.com/2023/freshman-survey/academi...)

[toasterlovin]

Legacy admissions actually make a lot of sense if you think that genetics affect the outcomes you care about but also that the relationship between genetics and outcomes is stochastic and messy (which it is, as breeders in the 1800s knew even before the mechanism was understood).

[dragonwriter]

Well, no, if you think that legacy admissions are unnecessary (because, to the extent genetics have an effect on the outcomes you care about, they'll show up in more direct measurements), and counterproductive (because they presume a simple relationship rather than a stochastic and messy one.)

OTOH, legacy admissions make a lot of sense if the outcome you care about is serving an elite class defined rather simply by familial lineages.

[toasterlovin]

If you think genetics matter for success (or whatever you want to call it), test scores are a proxy, but a successful parent is a direct measurement of the trait.

[dragonwriter]

No, if "the relationship between genetics and outcomes is stochastic and messy" then having a successful parent is a proxy measure, too.

[only-one1701]

This assumes, among other dubious stuff, that the initial admission of the ur-student was merit based.

[toasterlovin]

I’m assuming legacy admissions apply mostly to children of notable alumni.