Hacker News Reader

Comment by Cheer2171 | Hacker News Reader

> Grading is not "stack ranking". when I give out a grade to a student, I don't care how the other students did.

This was not my experience in a large public US research university for undergrad and masters in math and CS. So many of my STEM classes had exams that were not calibrated at all to any "bar" of minimum expectations. Often the median grade for exams (out of 100) was 30-60. So the prof fit it to a normal distribution and then put arbitrary cutoffs for what counted as an A, B, C, D, and F.

Social science and humanities was often very different grading. I had a humanities prof tell me exactly what you said when I asked about the curve: there is no curve, I do not compare you to other students, I compare you to my own standards.

View on HN

One of the amusing things I encountered in grad school was taking a course where both grad students and undergraduates were allowed. The course was curved for the reasons you mentioned but what was interesting was that the grad students and undergraduates were curved separately. To protect the undergraduates, right? Actually as it turned out the grad students scored worse on average!

The courses that the motivated and smart grad students take are typically not available to undergrad students. I took a lot of grad courses in undergrad, and then as a grad students I never studied with any undergrads since I had already taken those courses.

Though that wouldn't apply for motivated and smart grad students that are doing grad studies in a different discipline than their undergrad.

But basically all of the undergrad students taking grad courses are motivated and smart.

Depends on the university. At mine in CS you could take literally any grad course you wanted so long as the professor didn't mind (and they had no reason to mind), even if you didn't have any of the prerequisites. I learnt a lot that way.

Speaking from my own grad student experience, there is very little incentive to have a good GPA. Research is everything

Of course the grad students scored worse, they are unlikely to be asked about their grades ever again in their life if they are doing well with research and networking.

Undergrads on the other hand, a bad grade means screened out of interviews.

In my most extreme case of this, I had a C.S. midterm where I nearly doubled the second highest score (90 vs 47). The cutoff for an A was around 40.

The only one of my humanities classes that had a curve was the "Introduction to logic" (which was offered by the philosophy department). It was a rather gentle curve, but the tests were also open-book.

One of my favorite finals had two parts. One was a normal test on the material. That was not graded on a curve.

The other was a more open ended set of problems. That was graded on a curve, and was a lot of fun.

But yeah, every other CS class I took was on a curve.

I recall certain intro CS courses being calibrated to be difficult enough to fit with the department’s (generally tough) grading guidelines without actually applying a strict curve to exam results after the fact, so as to encourage students to see it as a collaborative endeavor rather than a competitive one. That seems kind of ideal if you can make it work. But yes, it’s hardly unusual for grades to work a lot like stack ranking.

Reply on news.ycombinator.com