Hacker News Reader

Comment by tsm | Hacker News Reader

As much as I would love to sit here carrying on about how Information Wants to be Free, I do appreciate where she's coming from. YBB forced to students to actually read course evals and apply critical thinking to their class selection. YBB+ wrongly suggests that the quality of a class can be reduced to a scalar value.

We've complained about this bitterly going the other way. How many app developers are frustrated when they get a one-star review that says, "This app couldn't sync to my online account unless I was connected to the internet. Changing to five stars when that's fixed"? How many Amazon shoppers have seen one-star reviews that say, "The package was damaged during shipping. The return was really easy and I got a replacement in two days, but I wanted it for an event that happened before the replacement came."?

View on HN

As an app developer who has gotten those reviews, the reality is they don't actually matter. Since everyone gets them, the real question is how many happy and 5 star reviews you get from the rest of your customers. You don't have 3 stars because of the guy that put 1 star, you have 3 stars because of the people who gave it 3, or all the people who didn't feel it was worth it to put down 5 stars.

The situation is the same here, at my school when viewing aggregate evaluation data, I find it to be VERY accurate to the actual experience I've had in those classes, and I think that's the issue here. Yale is protecting teachers from being called out for being bad teachers.

I feel like rating systems should be median based. Or bottom 25th percentile based. Those asshats that put 1 star "because it doesn't do x, otherwise 5" ruin 4 other 5 star ratings. When the bottom 25th percentile would still be 5/5.

I fully agree with you. For example when I look at Amazon reviews I never look at 5 star or 1 star review. Always 2,3,4 - those come from people that actually had issues - some smaller some larger. Then I can decide: ok, I might encounter these problems but it's something I can live with.

If the performance of the professor in a class cannot be reduced to a scalar value, then it seems probable that the performance of a student in a class also cannot be reduced to a scalar value, and professors should be required to write paragraph evaluations of students instead of just assigning them something from an A to an F.

In other words, if reduction of complex performance to a scalar value is invalid, then the entire grading system that defines academic success is bogus.

Incidentally, I think that likely.

Thank you. This is very insightful!!!

This is indeed exactly what happend at my lower/middle/high school--we received a full paragraph from each teacher evaluating our performance every trimester.

Doing so at a college scale would be too costly. But you do have all the grades that go into your average, comments on each essay, office hour time, etcetera. The nuances of your performance don't matter in your GPA, that is, your Grade Point Average, but you do have a bit more information available to you, yourself.

My alma mater (UC Santa Cruz) does exactly this, all course were pass/fail and if you passed you got a narrative evaluation, which ranged from terse sentences that obviously encoded traditional grades as text ("He did very good on the midterm, and excellent on the final, and excellent overall.") to some that were several paragraphs in length and discussed specific merits of the work done in the course. Amusingly they also had a committee that would translate your narrative evaluations into numeric grades so you could apply to grad schools that required a GPA.

Actually, my experience in college was that classes could usually be reduced to a scalar value.

Regardless of subject matter, I learned the most in classes that had good teachers who received good evals.

However, the administration doesn't think that way; they think that it's more valuable to take the "right" course than to take a course with the right teacher.

> YBB forced [] students to actually read course evals and apply critical thinking to their class selection.

This doesn't come close to being true in any sense. Wanting to do something and actually doing it are barely even related concepts.

There is already some data that Apple doesn't provide in the app store but that third party sites do collect, namely price history. Being able to see if an app often goes on sale is very useful to the user, but not information that Apple or the developers want to make obvious.

Reply on news.ycombinator.com