I feel like you didn't read what I said, but rather disagreed and then echo'd it with more words :P
When I say linear curving, I'm referring to a 0 slope, a raise by a constant percentage. That may have been where we flew off.
Anyways, the Profs tend to curve the class average to some value, then apply the increase everyone's marks. That is to say, they curve a variable amount. So in a class of say 20 students, the median was 60, but the average was 50 then indeed the lower end marks get affected the most. However, if the average was say 60% and a median of 45% (kinda drastic, but just an example), then you'd notice that the curve applied to the class average is very very small. In the end the guy who scored 10% in this case can (probably, depending on the distribution) not hope to get over 20%. That was probably the ♥♥♥♥tiest explanation ever, but you get the idea.
I've taken calculus pretty much every year so far. I've studied from integration to real analysis and number theory for my program requirements (even done some linear algebra stuff, design, analysis of data, probability & statistics) . This grade curving problem is more like applying a function to a set of grade distributions than it is calculus.