Well, character power as described in RG 2.21 work as you describe. And those are the rules we use for RoE 3. And thank goodness for that.
But Brandon and I were discussing 2.25 where the rules have been changed quite a lot.
Brandon thought that in light of the new proficiency rules(in 2.25), perhaps conveying a negative bonus for being unskilled would be in order. Since then there would be an actual difference between low unskilled and low expert. I say that 2.25 don't improve by having successful support actions by AA convey a penalty to the action they try to help.
Perhaps he hadn't read all through the limited 2.25 rules and just wanted unskilled AA's to have less chance of succeeding their support..
I don't really know what he wanted to do, but I do know that as he proposed it he asked for the first rather than the latter.
In 2.25 it is in fact just as easy for a low level unskilled to support an action as it is for an epic master.. The difference is how large a bonus they give to the action when they succeed.
And when that is true then he should rather focus on the level of bonus the AA give if he want a difference between unskilled and expert.
I would like to point out that, you're front loaded in d&d rules..
Let us for instance say you are an expert at spellcraft. You could have a feat giving you +3, your int could be 16 giving you further +3, and you could have 4-6 ranks in the skill dependent on level giving you 10-12.
A master could further more have edges and flaws giving +2, a regional feat giving another +1 and 20 in intelligence for a total of 15-17 at low level.
The difference between a level 1 master and a level 10 master could be as low as +9, where as the difference between a level 1 master and untrained could be +15.
So if you wanted a system that were representative for how good a character could get in d&d 3.5, you should start them off with something depending on mastery level and then add a flat bonus to that for level abstraction, regardless of what mastery level the character has. (As I mostly did)
Lastly I would like to argue that my table is more linear than Brandon's in several ways.
His unskilled have an increment that is off compared to his other skills, that I think is vastly more different than the rest of his table than my choice to make epic master +10.