Here is where I will make lolsmileyface happy.
I think mechanically (meaning the process at which a judge produces a score) judging should be done on a base 10 scale (meaning out of 10, or 1, or 100). Why? Everything we do is based on base 10 so it is very natural to everyone. Second, I think that execution and difficulty mechanically speaking should be the same score. So if you judge difficulty out of 10 you judge difficulty out of 10. This way a judge could never accidentally give a wrong value because of mechanics. NCA's scoring system works the best for this because it is a rubric where you have to perform and compete certain skills to get into the right difficulty category. From there judging difficulty and execution is done in the exact same format. Execution, however, is done a bit different. Everyone starts with a .5. If you perform with an average ability you get that .5. If you are worse it lowers your score. Better, it raises it. Difficulty is a score that is always relative to your peers, and that is fine. As long as we dont value each thing in a routine individually (which would be very difficult and time consuming) that set range does give the judges a decent way to separate difficulty.
One of the many problems with other scoring situations is it seems to be out of a random number (20, or 10, or 35). That so goes against how we run our everyday life. You ask how attractive someone is, and they give you a score based off of 1 - 10. You ask how good is that mean, you say on a scale from 1 - 10 how was it? Who says on a scale of 1 - 35 what would you rate that? It goes against us.