I generally agree with you that grades should reflect the level of mastery of the material rather than some 'rank order' within a particular class. For that reason, I have always thought that grading on a curve did a disservice to both the students and those who had to try to make sense of the grades.
For example, in curve grading, two teachers teaching an identical subject should come out with essentially similar grades -- even if in one class almost everyone demonstrated substantial mastery of the material and in the other the range was from abysmal ignorance of the material to mediocre mastery of the material. In such a world, how could you ever compare grades between different sections of the same course, or between different curricula (e.g. engineering vs. sociology), or between colleges. Clearly, you could not. All an "A" would mean was that you were among the better students in the class. But, "better" doesn't necessarily mean good, or even adequate.
Therefore, if grades have any useful meaning at all, it is to distinguish the level of mastery of the material taught. There will always be material that is easier for some people than others, and always material that is generally agreed to be hard (e.g. advanced mathematics) and material that is generally agreed to be easy (e.g. 'communications').
Amen. Go, Cato!!