This may be irrelevant to the discussion, but my father wrote a thick volume entitled “Simple and Complex Intergranular Diffusion of Cobalt and Nickel into Mild Steel as Determined by Radioactive and Chemical Analysis” as his PhD thesis. (I didn’t make that up, it’s real.)
When it came time for the “final”, there were no written questions, the thesis wasn’t evaluated, he sat with a bunch of experts in metallurgy at the University who quizzed him orally on the contents of the document. Yes, it must have taken them some time to digest it and come up with relevant questions, but that’s how it went. He passed - with some sweating - and got the degree.
Flash forward about 40 years and he taught a graduate level once-a-week course at Stevens Institute in NJ. While there were infrequent quizzes along the way, the final exam always consisted of specific questions related to the thesis the students did throughout the semester, which went ungraded.
It was immediately apparent which students had done the work and which had bought somebody else’s paper from a fraternity file cabinet and submitted it. More than once a student looked at the questions, folded his booklet and left the room.
It took Dad longer to write individual final exams, but that’s what he did, I guess based on his experience with his PhD. Back in the days before the internet and AI, when “file cabinets in the fraternity” were all we had, this was a pretty sure method of fraud detection. Probably would still work today, if professors would spend the time.