Hacker News Reader

Comment by gambiting | Hacker News Reader

The issue isn't and never was with grading. No one is giving higher grades than they should in those UK universities that I worked with - the issue is specifically with lecturers being instructed to write exams in such a way that most people pass, because doing otherwise is bad for the department and for publicity. Which results in exams which are stupidly easy for anything that doesn't have "engineering" in the name.

View on HN

I envisaged creating exams would come under the remit of an examination board just as it does for A levels and GCSEs.

I just can't imagine how that would work? Maybe for some "core" modules - sure. But if you have a Russel group university with an industry leading ML course, where the professor behind the course is the leading researcher in the area - who is going to write the exam for it?

The "problem" here is standardization(or lack of it) - and it's actually very good that university courses are not standardized.

You just paper swap with another uni. Oxford can hand it's engineering paper to Cambridge and see if it's hard enough etc. Not hard to do, it's just peer review for exam papers

Exam difficulty can't be assessed solely on the paper, the difficulty is also a function of the curriculum - if they teach to the exam, covering only a few topics which are then examined, then the exam is easier than if it selects questions from a broader curriculum.

It's not just one board though, which shifts the problem, then the boards are motivated to make their exams easier because that wins them business from schools who are motivated to get higher grades regardless of the educational outcomes.

Reply on news.ycombinator.com