One of my professional classes, the class average was often in the 40's & 50's, and the professor would just curve it. I don't remember my individual scores, but I often hovered around the average in the class, so I probably got some 40ish, maybe even a 30ish score in that class. Personally, I think the professor has failed at his job when the "highest" grade in the class, is below failing.
It depends. I actually have the philosophy that in the graduate classes, exams need to reflect perfect understanding of the subject (and I mean perfect). The syllabus I have for the clinical trial database design (Piantadosi text and the Date temporal book) class quotes the A at the 45% and higher range with me having the discretion to move the curve in their favor. When they score 45%, that to me actually means that they know 45% of the material, and I'm happy with that. (And yes, even with that standard, I have years where someone scores in the 80's while the A's were in the 40's, those outperformers, I recommend they consider that as a possible career research decision.) Then again, I'm also the one of the most reviled graduate examiners because I send take-home exams with open book, open notes, and open conversation (as in, I don't care if they use the internet or post their actual problems because they are hard enough that crowds won't solve them), and they can submit exams jointly if they are taking the class together and they would receive the same grade. These sort of problems are ones of depth, not breadth, and part of the class's charm is to adapt the students' mentalities to the idea that the buck stops with them on these regulatory issues.
It's trivially easy for an instructor to set up students to fail, but it takes a good instructor to set up students to pass, and it takes a great instructor to push the limits of what that student talent can do. I've been very confident that I can draw out that potential from enough students (and they agree). That said, I've never thought that the sadistic approach works in the long run, but setting clear expectations that a class A is only close to half of what the subject contains is somewhat humbling. I find it personally challenging to set a difficult but fair standard as the material keeps on expanding.
For an undergraduate class where the knowledge base is supposed to be at a practitioner level, yeah, I can definitely agree with your sentiment that unless the instructor has explained the pedagogy of hard exams to knowledge base, s/he is just being sadistic and arbitrary. Supposedly, there is advice out there that what you do at work when you're not working should really be your work, but I often wonder why some of my colleagues would choose education when their proclivities are toward sadism rather than success.