Curriculum design and assessment design is not a new field, it is just an oft ignored field. The first step is to make very explicit goals. The downside of explicit goals is failure of the institution itself is potentially measurable. "Coverage" is a conveniently wishy-washy standard that makes it easy to pass the buck to the individual students -- "we covered that" has the convenient ring of authority.

Curving the grades is an awful idea, but how it is usually applied it is rarely harmful. Professors mostly use it defensively in favor of the student, to fudge the grades up when the test turns out to be harder on the students than the professor anticipated. I am sure an occasional professor does the opposite, but one can never be truly safe from terrible instructors (other than avoid them).

I think it is entirely possible to use a curve, or even a scale to adjust a percentage score to a numerical grade. You can establish a minimum acceptable proficiency... say 55%, less than that, you fail. over that, you can have a curve that adjusts against each student and other students in that class (or even prior classes) on the same level/exam for a final grade... accepting 10% will get a D, and 10% will get an A, with the rest spread evenly with a B and C.

> I think that tests and teaching in general could use a lot more discussion and analysis

You're right, but so many times when this is brought up, the result is one of those "express your feelings on this topic" type assignments. A few of those bad apples ruin it for everyone.