AEmer wrote:So my claim is the following:
We can actually study the concept of merit, and how to best assign merrit.
The current system is as follows: Universities have a reputation to uphold. If they put out crappy diplomas, their esteem will (theoretically) decrease, and people will find them less attractive. If they put out good diplomas, their esteem will (theoretically) increase. Therefore, they're incentiviced towards making a grading scheme which isn't crappy, but accurately measures peoples merit.
So that's the theory. But in actual fact, universities just need to uphold their reputation in any way they can; so long as they are well published, have charismatic professors and scientists, and have a strong media profile, the reputation would probably be pretty hard to ruin. Put on high entrance requirements, and the majority of the merit assesment is not even in anything tested at university, but comes from a cursory reading of previous grades and the hardship of attending uni at all: you can only pass if you get in, and you can only pass if you have the means to go through the years spent in education not doing anything that earns you money.
So just with these facts alone, it becomes clear that the incentives universities have to get merit right aren't that big. In fact, just being expensive and prestigious, and using some arbitrary fashion to ensure that a certain number of people always fail, combined with headhunting professors and having high entrance requirements would probably do about as well as many current universities. With this kind of sorting and a large enough battery of exams and presenting relevant subject material, the graduating students will probably be decent enough that everything will keep running smoothly, even if merit is assigned very unfairly among any given group of students.
So we have two alternatives: We can either presume that this system is as good as it gets, or we can presume that it is not as good as it gets. I presume that it is not as good as it gets because the universities who design the merit system do not have that much rational self-interest to getting it right. My argument is that it's not optimal because nobody really counts on it being optimal: So long as the job is done, the system is able to sustain itself.
I'd argue that almost regardless of how tricky it is to develop better merit assesment, doing some amount of dedicated work including actual simulations and studies and experiments would help, because with such at least _someone_ would have an incentive towards understanding the problem and getting it right.
Interesting thought. There is yet another issue: government funding. In the Netherlands, universities get money for every diploma they hand out. If a student graduates in three years, that's good, if that student takes 6 years, that's bad, because you're spending twice as much resources and get only the same amount of funding. This obviously creates unwanted incentives, although universities are already regulated and monitored on quality and what their programmes should cover.
I think the current system (at least here in NL) is quite decent: those who pass are generally significantly more knowledgeable than those who do not. Those who score 9/10 are not necessarily better than those who score 7/10: some people are quite smart but are simply not suited for doing tests in general or timed tests or multiple choice tests &c. However, in general those who score higher grades tend to be the brighter and more knowledgeable students.