Since these are comparative scores, they could only be given after the fact. How do you keep track in large fields?
Here is the best (in the opinion of 5 of the judges, IIRC) men's short program from 1996 Worlds: http://www.youtube.com/watch?v=IyGkx7b_Dc0
It was also the 2nd out of 30 performances that day.
What kind of notes do the judges need to keep to remind them to give 3 or 4 hours later that that was the best performance of the day? They can't give the best-in-field score in real time, because for all they knew that could end up being only be the 4th or 5th best performance of the day.
How do they remember that the first skater of the day was, in their opinion, 25th best overall? That's what the placeholder scores were for under 6.0, with some advantages from the flexibility of tiebreakers and some disadvantages in that the numbers corresponded only roughly to absolute skill levels as understood given the state of the sport at that point in skating history.
What the program components scoring is trying to do is to peg the numbers to those somewhat-less-rough mental consensuses about "absolute" skill level rather than comparing them directly to other skaters in the same event.
So I have a few questions about how your proposal would work.
*How many numbers does each judge give to each performance?
*How do they keep track of comparisons among many skaters across the several-hour duration of a large event?