Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Reinstating our SAT/ACT requirement for future admissions cycles

mitadmissions.org

Post reply on HN
Print view View original

81–90 of 1001 posts

Re: Reinstating our SAT/ACT requirement for future admissions cycles

#81

Post by dontreact »

rossdavidh wrote: post #48
Animats wrote: post #20

All MIT students, regardless of intended major, must pass two semesters of calculus, plus two semesters of calculus-based physics, as part of our General Institute Requirements. ... There is no path through MIT that does not rest on a rigorous foundation in mathematics, and we need to be sure our students are ready for that as soon as they arrive. For a period in the 1980s-1990s, you could argue that calculus was not…

True, but the usefulness of calculus to a working developer in ML is pretty marginal. The difference between different ML algorithms for hill climbing or gradients, is orders of magnitude less than the effect of having the right training data, formatted the right way. Statistics or data science is far more applicable to nearly any field on real-world programming. But, you know, like Latin in the 19th century was alwa…

I disagree with this. Being able to read and understand the math in the paper of the algorithm you are implementing is useful, and calculus is common in ML papers.

You may have some short term success without understanding the algorithms at all, but as the field changes and you are no longer in school, being able to keep up at least somewhat with papers is very useful.

I agree that the day to day is mostly about formatting data though!

Re: Reinstating our SAT/ACT requirement for future admissions cycles

#82

Post by mason55 »

satsuma wrote: post #42
paulmd wrote: post #27

Earlier quoted context omitted.

yeah this is a really hard problem and I don't see a fix outside of standardized testing. Every school is individually incentivized to use every trick - grading out of 5.0, grading loosely, giving a bonus score for AP/IB courses "because of difficulty", etc and teachers are obviously very sympathetic to the future of their students and the impact that being a Grading Nazi could have. And parents are obviously incenti…

i came from a rural high school that didn't have any ap/ib courses. i wonder how much that affected my college applications. i still got to go to the college of my choice (fire up chips!) but i have to wonder -- if i was able to boost my gpa using ap/ib courses, would i have received more scholarship opportunities/better offers from other schools?

That assumes the AP classes would have boosted your GPA. If the harder class knocked you from an A to a B then it would have been a net negative, at least at my high school (AP counted as a 1.2 weighting, so an A in a regular class is 4.0 and a B in an AP class is 3.0 * 1.2 = 3.6).

Re: Reinstating our SAT/ACT requirement for future admissions cycles

#83

Post by Loughla »

thescriptkiddie wrote: post #69
LewisVerstappen wrote: post #7

Earlier quoted context omitted.

It's far easier for rich students to game GPA, college essays and extracurriculars than it is for them to game the SAT. For GPA, they can hire a private tutor in the subject they're struggling with. There are services out there that will basically write your english essay for you / do your math/science homework. For college essays, they can hire college counselors to help them draft a compelling essay. For extra curr…

I wonder how much taking the SAT multiple times plays a role. Personally I took it at least three times, and my score improved each time (don't remember by how much).

Which is also tied to socioeconomic status. If you can pay, your score will go up the more you take it, according to my experience, and the college board.

If you can't pay, you take it the one time it's offered for free if your school offers that. Then you get what you get.

Re: Reinstating our SAT/ACT requirement for future admissions cycles

#85

Post by hhs »

likpok wrote: post #71
sjg007 wrote: post #49

Earlier quoted context omitted.

Coaching and studying improves SAT scores. People learn the type of questions they do poorly on and can study to improve. The SAT is a test you can study for.

The top link is a study that shows that the improvement is pretty marginal in practice. It does miss something: in specific ethnic enclaves SAT coaching is much more effective, perhaps because of a culture of out-of-school schoolwork and teaching beyond SAT prep. Those enclaves aren't particularly wealthy either (if I recall correctly it was a Korean enclave). Even then we're talking 70 points -- not nothing, but als…

> Even then we're talking 70 points -- not nothing, but also not a radical transformation.

That would depend on the baseline score. For instance, if it was a 1510 baseline and then went up +70, then it would be useful.

Re: Reinstating our SAT/ACT requirement for future admissions cycles

#86

Post by School-Cotton »

paulpauper wrote: post #2

our ability to accurately predict student academic success at MIT 02 Our research shows this predictive validity holds even when you control for socioeconomic factors that correlate with testing. It also shows that good grades in high school do not themselves necessarily translate to academic success at MIT if you cannot account for testing. Of course, we can never be fully certain how any given applicant will do: we…

> So much for that common, popular notion that standardized tests do not predict anything of value.

It was always an example of people refusing to believe something because it would be nicer if it weren't true.

Re: Reinstating our SAT/ACT requirement for future admissions cycles

#87

Post by JamesBarney »

globuous wrote: post #4
paulpauper wrote: post #2

our ability to accurately predict student academic success at MIT 02 Our research shows this predictive validity holds even when you control for socioeconomic factors that correlate with testing. It also shows that good grades in high school do not themselves necessarily translate to academic success at MIT if you cannot account for testing. Of course, we can never be fully certain how any given applicant will do: we…

What's surprising though, is that APs and similar exams are not enough. In the UK, I though they essentially looked at A Level results, which are much more representative of what you'll actually study at uni. But I guess both SAT/ACTs & APs must be a better measure that just APs. I just remember fucking hating studying for the SATs though. So boring. SAT IIs were somewhat fun to study for though. In France for instan…

There are a lot of highschools where AP classes aren't really available, or are taught with varying degrees of rigor.

Re: Reinstating our SAT/ACT requirement for future admissions cycles

#88

Post by __MatrixMan__ »

LewisVerstappen wrote: post #7
paulpauper wrote: post #2

our ability to accurately predict student academic success at MIT 02 Our research shows this predictive validity holds even when you control for socioeconomic factors that correlate with testing. It also shows that good grades in high school do not themselves necessarily translate to academic success at MIT if you cannot account for testing. Of course, we can never be fully certain how any given applicant will do: we…

It's far easier for rich students to game GPA, college essays and extracurriculars than it is for them to game the SAT. For GPA, they can hire a private tutor in the subject they're struggling with. There are services out there that will basically write your english essay for you / do your math/science homework. For college essays, they can hire college counselors to help them draft a compelling essay. For extra curr…

I used to take the ACT for people, they'd pay me in beer (which I wasn't old enough to buy for myself). You'd get ID'd at the entrance, but nobody kept track of whether the name on your test was the name on your ID, so you'd just take each other's tests.

It might be harder now, I don't know.

Re: Reinstating our SAT/ACT requirement for future admissions cycles

#90

Post by uwuemu »

Animats wrote: post #20

All MIT students, regardless of intended major, must pass two semesters of calculus, plus two semesters of calculus-based physics, as part of our General Institute Requirements. ... There is no path through MIT that does not rest on a rigorous foundation in mathematics, and we need to be sure our students are ready for that as soon as they arrive. For a period in the 1980s-1990s, you could argue that calculus was not…

I would argue exactly otherwise... with the advent of the web platform and frameworks like .net, the vast majority (and I mean like 95%) of developers will never touch anything ML related in their careers. I mean, I get that this is MIT and many of their students will end up working with ML, but applying that globally to CS is nonsense. Back when I was studying CS (cze) more than a decade ago, we had to pass linear algebra, graph theory and calculus, but honestly, that was like in the first year and a half and then it completely tapered off (later years were all about projects, algorithmization, i.e. "doing the work" and very little about hardcore theory) and guess what, I never needed it again. A bit of statistics and some graph theory here and there, but that's about it.

Contrary to popular belief, there are NOT that many ML jobs out there and the ones that are there are more about data science and messing with model zoo type of shit than actually coding useful programs. Most programmers will be lucky if they get to integrate inference of a prepared model into the apps they work on.

Post reply on HN