In a science or engineering class, it might be everything above 40%.
A test which has a significant fraction of scores at the right side of the distribution has saturated the range: there are differences in comprehension and mastery between one 100% and another, which aren't being measured by the test.
The platonic form of a perfect test would have exactly one student per class who got every question correct, but that's impossible to reach except by luck. So a teacher is stuck between writing a test which is too difficult for everyone, and one which is too easy for a substantial fraction of the class. They should pick the first over the second.
While that can feel frustrating (I know, I've been there!) it's likely that you in fact learned the material better than you would have in a class where your level of mastery allowed you to score perfectly for an A+ and in the mid 80s for a B.
No, the teacher is not stuck here. It is rarely the purpose of an academic test to plot students on a wide distribution. That’s just something fun for teachers to do to look for exceptional students.
If the purpose of a test is to determine if an individual student passes some bar of understanding, then there is absolutely no reason to make it extremely difficult and then give A’s to those who got >50% correct.
It shows you what the next level is, reminds you that there's more depth to the subject than you can master in 12 weeks.
As long as you get the letter grade you deserve in the end, why not have an interesting challenge along the way?
Because of its wide applicability to business environments overall, the course was mandatory for some non-CS students. And boy, they hated it.
EDIT: Or the imposter gave the interview and added a mark up ;)
So, we've slowly grown from 80%->90%->95%.