When I get papers back for class, there's often a question or two that I know I could make a case for getting credit for my answer, but ultimately decide it's not worth the extra % in the grand scheme of the assignment. (e.g., on an 80 point midterm, I may not bother questioning why I got a 2 point question wrong).
My rationale is that the weight of this question is very small (2.5%), and not worth the effort really, when considering this actually translates to 2.5% of 40% of my final grade.
However, what if the whole class does poorly and the exam is curved heavily? At that point, does it make more sense for me to argue my 2 points? With the curve (assuming traditional curve), would it effectively make each question worth more at that point (since a curve could be seen as being given some points for free, per question)? So, would I actually be arguing for my 2 points plus the extra points that on inherent in the curve for this question?
This seems like the answer is "yes", but without some sort of formulaic proof, it's hard for me to convince myself.
Yes, it can. It depends a bit on how the exam is rescaled. Let's assume your teacher uses a very simple scaling by linearly taking the best score to $100\%$ and the worst score to $0\%$. Let us also assume that you have gotten $70$ and everybody else in the class has gotten $71$. This is clearly a terribly written test and an extreme example. For now, you have $0\%$ on the exam as you have the worst grade. If you get $2$ more points, you will have $100\%$ on this exam.
This is extreme, but the basic concept is that scaling that spreads the scores will make those $2$ points more valuable. Scaling that just shifts the scores will not change the value of the $2$ points. There may be scaling at the level of the total score, which again could make the points more valuable.