The solution of course is not to prevent cheating, but to make it irrelevant. We need to find new ways to assess the student performance, such that cheating will be unnecessary or impractical, i.e., not worth it. There are ways for small classes (project based assessment), but these do not scale up easily.
I think I don't recall ever being tasked to answer questions set in a textbook at university. The lecturers set problem sheets themselves based on their lectures. Sure, you can still cheat from other people in the same lectures, but that's a different game to having cheating commercially available.
That said, the exams were pass/fail anyway and the questions worked on during the course didn't have any bearing on that. Made cheating on problem sets pretty irrelevant, I guess.
>There are ways for small classes (project based assessment), but these do not scale up easily.
yes, project based learning, even in groups, can make cheating irrelevant. and it can scale if you (the lecturer) put in the effort to design it that way. its possible to do this with hundreds of undergraduates if you provide clear guidelines and assignment examples (i used to have a mythical student who we 'watched' complete her assignment). you can also make available an archive of all past assignments (and shift the topics each semester). in short, i used to encourage my students to "cheat", but the project tasks were always (interestingly) different.
What I don't get is why nobody seems to be willing to blame the professors.
Why is it so unreasonable to suggest that professors might create their own problem sets and tests? Solution manuals and test banks have freely existed on the internet for years. Every semester when they start their new classes, they know whether or not their students can cheat on the problem sets and tests that they plan to assign, but they throw up their hands and claim that if cheating is to occur, the students are to blame for lacking integrity.
Professors should look at the incentives for students and the stakes involved (GPA -> internship, GPA -> grad school, GPA -> future earnings) and realize that if there's an opportunity to cheat, some percentage of students are going to take it. And in fact, it's probably a wise decision. Professors like to say, "If you're cheating, you're only hurting yourself," but that's total bullshit. If you're cheating, you're hurting other people. In a world where classes are curved, it's a zero sum game, and from a game-theory perspective it might make sense to cheat even if you otherwise wouldn't. Once 20%-30% of the class is cheating, you're really just hurting yourself if you're not cheating.
I never cheated in school, and that's a nice little feather in my cap, but if I could do it again, I probably would have.
> Why is it so unreasonable to suggest that professors might create their own problem sets and tests?
Because creating a problem set or test that 1) is pedagogically useful and 2) solvable in a defined amount of time is a very difficult problem and certainly takes up a lot of time.
My solution to Chegg is to make assignments and tests that you don't need to answer every question for and put a couple REALLY difficult problems on things--technically solvable by the class by effectively not due to complexity/time. In a CS class, it's really easy to create simple looking problems that spiral out of control. I've occasionally done it even when I didn't mean to. :(
Anyone submitting a correct answer to it gets my detailed attention for cheating.
In addition, in a technical track, I do point out that not learning my class, which is a prerequisite, will absolutely kill you in the next class.
Between the fact that the universities that I have worked with will and have expelled people for cheating and the fact that my course is a prerequisite, I seem to have dissuaded the cheaters to take other professors. :)
Alternatively, they're just that good at cheating. Fortunately, the students who need to cheat are generally not smart enough to cheat well, either.
If you create a metric, it will get gamified until it's no longer measuring anything.
Especially now with remote school, where students pay regular tuition to get the privilege of having multiple choice exams proctored via some spyware, I can see why some just see it as an exercise to be min-maxed.
Caltech always had a strong culture of take-home exams and academic honesty, so it's really a solved problem.
I think Caltech's solution may depend on only accepting a small subset of the population. If that's the case, it's obviously not practical for most universities.
The most interesting part of the article, to me, is the estimate that 2/3 of college students cheat. I think it's surprisingly plausible. Personally, in one algorithms class I took, I was tasked with peer-grading ten other students' homework, and at least eight of them were clearly copied from the solution.
People optimize for the metric of intelligence not actual performance, reducing time and difficulty. The education industry is not about making people smarter, just getting them to grind away at problems and measuring their performance:
like a grind-centric video game, except a thousand times more boring and of course prone to automation.
