The hundreds digit is the amount you bump your success level up by, the rest is the number you're rolling against. You just look at your skill and read the last 2 digits. 133% skill? Crit on 33% or less, bump by one. 257% skill ? Crit on 57% or less, bump by 2. No extra calculations required. The numbers are right there in front of you.
I think that still introduces some sharp changes along the skill range though. I assume a "bump" means you increase the success level by one automatically, right?
I've got a few observations on the skill process in general. First off, I think the OP started by talking about non-combat skills (opposed skill checks) but the discussion has evolved into one about combat skills. Nothing wrong with that, but the two types are different (and are treated differently in RQ2/3 as well as in the new rules).
Frankly I never had a problem with the opposed skill process that RQ3 used. We use a system where you simply make your roll, then the opposed roller attempts his skill but sutracts from his skill the amount you succeeded by. It's a bit more math, but not too hard. Most importantly, it doesn't taint base failure/success values significantly (as long as you judge success by how much you make your skill by instead of crits/specials, this works perfectly).
I think the key concept here has been touched upon by several people. Opposed rolls aren't so much about your total actual skill as how much more or less your skill is then the other guy's. This makes the process work identically (in principle) to resistance rolls, and IMO conceptual consistency is a good thing in an RPG. This process lends itself very nicely to the standard "reduce/increase skills by X amount in Y situation" mechanic, since by not caring about the actual skill value but rather its numerical advantage on the opposed skill (or GM set difficulty if you will), all effects have equal "weight" regardless of total skill value (but may not be significant at all if the character is sufficiently skilled, which is the point).
We use a similar process for combat skills in RQ3 (stolen from RQ2 of course!), where you allow the character to subtract skill over 100% from the opponent's opposing combat skill. Once again, the point isn't relative skill as a percentage, but value +/-. The thinking is that skills are not "relative" past 100%. It's a sliding scale. A 200% warrior is not twice as skilled as a 100% warrior. He's simply 100% better. By making the skill subtraction optional, you give the player the choice of either an increased crit/special chance *or* an increased chance of making his opponent miss his skill (or he can divvy it up if he wants, but I've rarely seen anyone do this). We also allow this only for natural skill. Bonuses from bladesharp/parry don't count (but can certainly protect someone's skill from being dropped too much). So magical power can make up for a lack of skill, but wont give you an advantage over a more skilled foe. We also have a house rule that you can only get a skill increase if the opposing skill is within 100% of yours, which further solidifies the +/- dynamic, and ensures that only fights within that relevant skill range (100% +/-) have any value in terms of experience.
Dunno. It works pretty well. It's simple. It scales well. Higher power groups will fight more skilled foes. As long as the GM scales the opponents (which he should be doing, right?), there's no problem. And certainly, those high skill characters can cream joe random wimpy guy, but they should be able to, and they gain nothing for it except in context of the scenario. Most importantly, you have a smooth skill progression where it matters (opponents somewhat near your own skill level). IMO, if a game system has to choose between accurately representing skill comparisons when skill levels are close or when skill levels are very far apart, I'll take the former every time. Unless you regularly run your characters into scenario significant opponents who are outside that +/- 100% range of course. But if you do that, I think your campaign will have other problems that no amount of rule changes will fix.
Hah. One last thing. Maybe I'm imagining things, but I have this sneaking suspicion that the "Price is Right"(tm) style "high roll under your skill" concept is a holdover from someone who liked the similar process that Pendragon had. I just noticed that the idea cropped up once in RQ3 (the dragon magic "don armor" skill), and seems to have gained ground in the dev for this release. I've never liked the premise. First off, it tosses the normal "direction" of skill rolls out the window. Secondly, it just doesn't scale at all. Far easier/better to calculate success based on the amount someone rolls under
their skill. That way, skill levels over 100% actually work the same as those under 100% (simple subtraction), and skill modifiers from difficulty/environment work seamlessly. It just seemed silly to me that a game mechanic could, for instance, prevent a noble dragonnewt from ever having more then 9 points of armor no matter how skilled he was, and the amount was totally random (again no matter how skilled he was). Abilities should be *less* random at high levels, not more. Yeah. Esoteric issue, but IMO relevant to game design. Some ideas sound neat, but really don't work in practice, and I've always felt that was one of them.