News for hackers. Stuff that matters.
Re:Top colleges should select randomly from a pool of ‘good ... by ForHackernews on Monday August 24, 2020 @05:43AM (#24258985)
You're assuming that standardized testing procedures, high school grading, and college admissions officers have the precision necessary to correctly distinguish between the 99.9997% perfect candidate and the 99.9998% perfect candidate.

All I can say is that's nuts. I got a perfect score on my ACT, and I know for a fact that I guessed on some questions. Am I really a stronger applicant than somebody else who made less-lucky guesses?

There's a garbage-in/garbage-out problem here where people who desperately want "meritocracy" are deluding themselves into believing they can accurately identify "merit" down to some extremely fine precision.

Re:Top colleges should select randomly from a pool of ‘good ... by commandlinefan on Monday August 24, 2020 @04:46PM (#24265323)
> Am I really a stronger applicant than somebody else who made less-lucky guesses?

Well, what you seem to be saying is that the selection process is basically randomized among the top candidates already, per the linked article... so, no change needed?

Re:Top colleges should select randomly from a pool of ‘good ... by samatman on Monday August 24, 2020 @07:23PM (#24266616)
It's actually an argument for making the SAT harder.

You can see on the Asian curve for the SAT, that cohort saturates the high end. The SAT should be so difficult that no one gets a perfect score without being lucky.

And lucky guesses on an entrance exam is a better filter than a lottery, because it feels like merit. For an MIT degree to mean "I'm reasonably smart and very lucky" would be a disaster, for society and for MIT students both. The illusion of agency simply must be maintained.

Re:Top colleges should select randomly from a pool of ‘good ... by unishark on Monday August 24, 2020 @06:37AM (#24259297)
On average people with slightly-higher scores/grades/etc should still be smarter.

Anyway the alternative proposed here is replacing lucky question guesses with luck via a lottery.

Re:Top colleges should select randomly from a pool of ‘good ... by ForHackernews on Monday August 24, 2020 @06:54AM (#24259396)
> Anyway the alternative proposed here is replacing lucky question guesses with luck via a lottery.

Yes, this proposal is more transparent and honest about the factor luck plays in the outcomes. It doesn't try to launder luck through some test numbers and essays to pretend it's purely meritocratic.

Re:Top colleges should select randomly from a pool of ‘good ... by unishark on Monday August 24, 2020 @07:38AM (#24259666)
Downvoted over "luck laundering"...

Definitely time to quit get back to work.

Re:Top colleges should select randomly from a pool of ‘good ... by notacoward on Monday August 24, 2020 @11:25AM (#24261550)
I thought it was actually a good phrase, and GP's comment contributed far more to the conversation than yours. Please try to do better.
Re:Top colleges should select randomly from a pool of ‘good ... by ForHackernews on Monday August 24, 2020 @08:44AM (#24260049)
Don't need to work, I've already proven myself a very lucky person. ;)
Re:Top colleges should select randomly from a pool of ‘good ... by AstralStorm on Monday August 24, 2020 @06:21AM (#24259200)
Single sample estimation is never that good.

Even were the tests perfect, rich people and ones with additional free time can take more attempts with reduced cost to themselves and families.

Additionally, tests are extremely time boxed, while actual research work is much more lax. Tests also do not measure cooperation, a crucial thing in real life research and development. Speed is not necessarily quality, and we don't know how correlated it is.

Re:Top colleges should select randomly from a pool of ‘good ... by murgindrag on Monday August 24, 2020 @04:48PM (#24265331)
75 years of test validity research say we DO know how correlated tests are with every conceivable definition of "quality." Your comment suggests you just haven't bothered to look.

Multiple attempts on tests don't help very much. We have 75 years of reliability research to show that. It's not single sample estimation. It's a couple hours to collect A LOT of samples.

Of course, money for tutoring, better schools, and highly-educated parents help with test outcomes and are a barrier to socioeconomic mobility. The basic problem is, random selection aside, no one has proposed a fairer system. Read Animal Farm at some point to see what happens when you have revolutions against an unfair system, without proposing something better to take its place..

But my basic point is you're confusing things YOU don't know with things WE don't know. We know a lot about tests, their upsides, their downsides, and alternatives. It's not like there aren't scientific conferences on this stuff.

Re:Top colleges should select randomly from a pool of ‘good ... by rhino369 on Monday August 24, 2020 @06:42PM (#24266363)
And the scientific consensus says they aren't in the same galaxy as ranking 99.9997% to 99.9998%. SAT, for example, is better than grades are predicting first year GPA, but it's still not that great.
Re:Top colleges should select randomly from a pool of ‘good ... by murgindrag on Tuesday August 25, 2020 @08:13AM (#24270522)
My point wasn't about the level of validity. My point about making inane comments. OP claimed we don't know things which we DO know simply because OP hasn't bothered to look.

Your numbers, 99.9997% to 99.9998%, roughly ask whether we can distinguish the second-best student in America from the fourth-best with an SAT.

I'm not quite sure anyone is either using the SAT for that, or claiming it's a helpful tool for that. Indeed, at that level, it's not even clear what ranking even means. If you believe people are using the SAT this way, please point me in their direction.

Again, inane comments aren't helpful for advancing the discussion. We have specific facts and numbers to work from. Trump popularized the art of making up facts on the fly, but it's not one I recommend adopting more broadly.

Re:Top colleges should select randomly from a pool of ‘good ... by ForHackernews on Tuesday August 25, 2020 @12:05PM (#24272744)
Harvard, Yale, etc. absolutely are claiming that they can separate the 2000th "best" student who gets in from the 2001st student who doesn't. If they wanted to fill their freshman classes entirely with valedictorians with perfect SAT scores, they have the applicant pool to do that.

My assertion is that there's no way for them to accurately make distinctions that fine. Much more honest to set some bar of "you must be at least this proficient to perform academically at Yale" and then pull names out of a hat containing everyone who meets that standard.

As it stands, the current admissions regime looks designed to reify class privileges in America.

Re:Top colleges should select randomly from a pool of ‘good ... by murgindrag on Tuesday August 25, 2020 @05:18PM (#24276252)
Where do Harvard and Yale claim that?

This is thread reads like a Donald Trump speech. AstralStorm claims science doesn't exist, when there are whole conferences. rhino369 fakes up some numbers. And now, you come in making fake claims about Harvard and Yale admissions.

I mean, yes, no one is arguing that the admissions regime isn't there to reify class privilege, but if you're going to try to fix the problem, you can't start with falsehoods, half-truths, or sloppy thinking.