An idea that's gotten a lot of attention lately is "competency-based education", essentially the idea that instead of having courses of fixed time (e.g. a 15 week semester or 10 week quarter) you have shorter modules that students take and retake as needed, and they move on when they've achieved competence in whatever topic/skill/idea/etc. they are pursuing as part of their educational program. For a professional program like business or engineering it is probably pretty clear what it means to achieve competence at some particular skill. For many of the more skill-based aspects of science I think it also makes sense. For humanities, I assume that once somebody has, say, successfully read some list of writers and produced critiques that analyze specified aspects of the work in light of specified concepts, one would also achieve competence, and then move on to some other list of works and ideas.
It's not a bad concept. It's not entirely objectionable. But to the extent that the idea is based on a critique of a traditional course, I want to defend traditional courses from the critique.
The critique seems to be that in a traditional course a grade of (say) B means that you got most of it and did pretty well but didn't get all the way, which is fine, but you never know what the student was strong on and what the student was weak on. To the extent that the critique is rooted in "you never know..." my question is "Who?" Presumably the answer is "The person reading the transcript." Fine. In response, my next questions are "Who reads the transcript and what do they want to know?"
I've spent a lot of time interacting with people who hire physics graduates. To a large extent they don't read transcripts at all, and maybe that should give us some humility about our enterprise. But before we conclude that our transcripts thus need to be more information-dense in order to be more useful, let me observe something else:
The employers that I've interacted with seem to care (at most) about whether students took a lot of lab classes and used a few specific tools in those classes. Beyond that, they just assume that students will need to be trained.
And that should not be surprising in science and technology. Everything is highly specialized and rapidly changing. And every employer is working in a different niche, and hence needs people for a different niche. Knowing that a person is smart and capable of learning seems to matter more than the specifics because of how steep the on-the-job learning curve is, even under ideal circumstances.
Given that, knowing very specific things is less important than knowing that the student has done lab work and can learn quickly. In that case, the person who learned a whole lot in 15 weeks really is more valuable than the person who would need substantially more time to learn the same amount (which competency-based education would allow for).
So what a grade in a traditional class really tells is what happens when you throw a lot of challenges at a person in 15 weeks. The real question is whether the ability to surmount those challenges is predictive of ability to learn on the job.