There's nothing special about assigning points to each answer, but the overall grading scheme I use has a couple of twists to it. These leave the top scores with top scores and move some of the lower scores up.
I try to remove bad questions from the test. Of course, I try to do this before the test, but there are always some that are still there.
< 50% correct. One measurement of bad questions is to look at the percent of the students were able to answer it correctly. If this falls below 50%, I consider it a candidate for removal. The scores may be low because the question was not understandable, because it was too hard, or maybe because it wasn't even taught. Exceptions: If I think it was properly emphasized during the course, or was asked on a previous test, I don't eliminate it.
Cancellation credit. If a question is removed, the points for getting it right can cancel out any other questions missed in that section of the test.
I usually scale tests to the best "realistic" score. That means that I take the top score and use that as the measure of 100% -- the highest that anyone can realistically score.
Exceptions. Sometimes there's someone in class who knows everything, but is just taking it to get the credit, and who consequently gets near perfect scores on everything. Or maybe there's a genius in class. I try to eliminate these outliers if figuring a realistic maximum score somewhat lower.
Scaling. The difference between the realistic maximum and 100% is then either added to everyone's scores, or the numbers are scaled by multiplication. No score is allowed to exceed 100%.