Comment by leetcrew | Hacker News Reader

Comment by leetcrew | Hacker News Reader

Parent Full thread

> I think curves are in general unethical due to cheating, and feel they’re a sign that a professor hasn’t done the hard work to really zero in on exactly what knowledge the student is expected to master.

I disagree; there's no objective criteria for what students should be "expected to master" in a particular course. it's inherently relative to what the typical student at that institution is capable of. a class where everyone gets an A is probably a waste of time for everyone involved. it strongly implies that more material could have been covered.

if a whole institution is like this, it gets back to the original problem. when everyone else is graduating with a 4.0, a 3.8 looks a lot like a 2.0 from a more rigorous school.

ideally, the material itself would be designed to get a good distribution of As, Bs, and Cs with a few Ds and Fs for people who didn't try or understand at all. but it's pretty hard to get this exactly right. better to err on the side of making things a little too hard. then the occasional bright student will really shine, and you have enough signal to compress the range into the expected letters at the end.

there's no objective criteria for what students should be "expected to master" in a particular course

Not exactly, but depending on the course you can get pretty close. In my engineering statics and solids classes, it mapped well with what you'd be expected to do when working as a stress engineer (which is what I worked in after school). In my heat transfer course, it mapped well with the responsibilities of a thermal engineer.

ideally, the material itself would be designed to get a good distribution of As, Bs, and Cs with a few Ds and Fs for people who didn't try or understand at all. but it's pretty hard to get this exactly right. better to err on the side of making things a little too hard. then the occasional bright student will really shine, and you have enough signal to compress the range into the expected letters at the end.

And that's exactly what my professors who didn't curve managed to do. It was clear they worked very hard at prioritizing the important material, teaching it well, and testing it fairly. It was a breath of fresh air.

But when the average score on an exam in one of my other classes was 22% -- and they weren't looking for the next Einstein, it was just another upper division engineering class, presumably the geniuses would have revealed themselves by that point -- it was clear that the professor wasn't even trying. Throw a bunch of crap on the test and let the curve sort it out, so the professor could get back to what they really wanted to spend their time on: research.

Research also happens to be what they are evaluated on for tenure/promotion etc. Teaching evaluations are generally a tiny part of that equation.

> when everyone else is graduating with a 4.0, a 3.8 looks a lot like a 2.0 from a more rigorous school.

I've never heard of this interpretation before, it seems this is a difference in whether the GPA should represent a student's actual grade on assignments, or the student's overall achievement relative to their piers. It seems the curve exists for the latter ideology - you can't expect every FAANG recruiter to say "well they got 2.9 from Georgia Tech, that's better than this 3.4 from Duke", if they did, you'd probably have pretty arbitrary hirings (although, if it became policy, I can see some Googler making an internal tool to 'normalize' school GPAs); although it seems MIT has a "no curves" policy and graduates still manage top-tier GPAs.

I can't speak to what goes on in FAANG recruiting, but I was involved in hiring for a smaller company that recruited heavily from regional schools. we absolutely knew which schools were harder, as most of the younger engineers had graduated recently from that same set of schools. obviously GPA doesn't tell the whole story, and we preferred to decide based on work experience. but for junior hires and especially interns, there's not always a lot of signal to decide on. all things being equal, we would prefer someone with a 3.0 (or even lower, with a good explanation) from the rigorous stem school over someone with a 4.0 from the well-known party school. which is too bad, I'm sure there were some very bright people who went to the "party school", but their grading policy made it very difficult to distinguish them from their peers who barely had a pulse.

Reply on news.ycombinator.com