School ideally provides two things: domain expertise (by way of education) and the perception of domain expertise (by way of a credential). The problem is that these two functions are significantly disjoint. Studying for exams, doing research, and doing projects can provide education, but it is not necessarily so. Likewise, one can be educated effectively without actually completing such assignments.
As such, someone who is more interested in one half of the result can optimize to obtain that half with less effort than would be required for the whole. These are not likely to be brought into alignment any time soon. Session exams, if re-created from scratch and kept confidential, are difficult (though not impossible) to cheat. Unfortunately, they are typically possible to cram, still possible to cheat, and they reward test-taking skills as much if not more than subject mastery.
The practical exam is probably the most effective, where one purporting to have a skill is simply asked to demonstrate it on the spot. Unfortunately, not everything is amenable to this, and they are still possible to cheat, though there are really only two ways[1].
Practical exams are unfortunately not so good for research-heavy subjects (like, say, all of graduate studies). You can do it (research this topic, take all the notes you like, bring them into class, answer some questions (possibly in long-form), but it is a less effective exam than the normal research paper if the latter is done honestly. Still, I think they are the only way to go if you really want to put an end to cheating.
[1]-Impersonation and outright bribery. There was a big scandal about the former in one of my classes back at Uni, in which six students were expelled. I've never heard of the latter, but I find it implausible that it never occurs.
You can do it (research this topic, take all the notes you like, bring them into class, answer some questions (possibly in long-form), ...
At least some university do exactly this for the Comprehensive Exam required as one part of a master's degree. My wife is preparing for something very much like you described for her master's in history right now.
As such, someone who is more interested in one half of the result can optimize to obtain that half with less effort than would be required for the whole. These are not likely to be brought into alignment any time soon. Session exams, if re-created from scratch and kept confidential, are difficult (though not impossible) to cheat. Unfortunately, they are typically possible to cram, still possible to cheat, and they reward test-taking skills as much if not more than subject mastery.
The practical exam is probably the most effective, where one purporting to have a skill is simply asked to demonstrate it on the spot. Unfortunately, not everything is amenable to this, and they are still possible to cheat, though there are really only two ways[1].
Practical exams are unfortunately not so good for research-heavy subjects (like, say, all of graduate studies). You can do it (research this topic, take all the notes you like, bring them into class, answer some questions (possibly in long-form), but it is a less effective exam than the normal research paper if the latter is done honestly. Still, I think they are the only way to go if you really want to put an end to cheating.
[1]-Impersonation and outright bribery. There was a big scandal about the former in one of my classes back at Uni, in which six students were expelled. I've never heard of the latter, but I find it implausible that it never occurs.