Goblins & Greatswords: An alternate resolution mechanic for the thiefly arts and other skills

A while back, I posted my ideas on how the skills of thieves (plus a few others) would work in my fantasy heartbreaker, Goblins and Greatswords.  I chose to stick with the percentile dice model used by most old school iterations of D&D, as opposed to a simple d20 or d6 mechanic.  One reason for this is that rolling two dice allows for special results to be triggered on a roll of doubles.  Another is that the percentile dice "roll-under" format allows degrees of success to be easily determined and scale in the right direction: just take the tens digit of a successful roll, and the highest degrees of success are possible only with the highest levels of skill.

However, there are some things I don't like about it too.  It's very inelegant when applying modifiers, for one, and ability score modifiers are something I very much want to include.  The standard -3 to +3 ability score modifier is dwarfed by a 100-point range.  Sure, you can convert those to plus or minus 5, 10, or 15%, but you're still asking your players to crunch bigger numbers at the table, and either adding them to the base chance or subtracting them from the roll, which feels weird.  Then there's the problem of looking up numbers in a table every time you want to do something.  It's a lot easier to remember that you have a +5 bonus to your Stealth skill than it is to remember that you have, say, a 47% chance of success.

So, I'm considering a system using a roll of two dice, but adding them together in a roll-over format.  My first thought was 2d6, but the range just isn't big enough to accomodate both improving skill by level and modifiers.  2d10 has the range to work, but the "success" point would have to be at some wonky number like 16 in order to start with reasonable odds, and 2d8 has a similar issue. 

2d12, now...that's interesting.  (Go here and click on the "At Least" tab if you want to follow along with a visual aid.)  There's a 10.42% chance to roll 20 or higher, which means that, if you set the target number at 20 (intuitive and easy to remember!) the average schmuck who has no bonus in a skill would succeed roughly 10% of the time.  Start out with a +1 bonus, and you're up to 14.58%, which seems good enough for a dabbler in the skill.  A more serious student of a skill might start at +2, for a 19.44% chance, which maps pretty well to the beginning percentages of most thief skills in B/X. 

Add a bonus for a high ability score, and a character could start with 25%, 31.25%, or 38.19% odds - a meaningful bump, but not so much that it swamps the whole system.  There's still lots of room for improvement, which is desirable because I want leveling up to mean something, and it doesn't if you're bumping against the 100% success ceiling too soon.

Rather than using a table of percentages, increasing at different rates for Basic, Good, or Elite skill progression, I'd use a relatively simple formula:  Basic starts at +1, and gains an additional +1 at odd-numbered levels.  Good starts at +2 and gains +1 at levels divisible by 2 or 3; thus at level 2, 3, 4, 6, 8, 9, 10, 12, 14, 15, and so on.  Elite also starts at +2, and gains +1 at every level.  One interesting feature of this is that Good and Elite are essentially the same at beginning levels -- all the way through 4th, in fact -- but Elite slowly pulls ahead at higher levels. 

Here it is in table form, so you can clearly see the relative progressions.

Level	Basic	Good	Elite
1	+1	+2	+2
2	+1	+3	+3
3	+2	+4	+4
4	+2	+5	+5
5	+3	+5	+6
6	+3	+6	+7
7	+4	+6	+8
8	+4	+7	+9
9	+5	+8	+10
10	+5	+9	+11
11	+6	+9	+12
12	+6	+10	+13
13	+7	+10	+14
14	+7	+11	+15
15	+8	+12	+16

If levels top out around 15 (and really, there's not much reason to go beyond that, is there?) then a Basic skill ends up succeeding 61.81% of the time, Good 85.42%, and Elite 97.92%, before any ability adjustments.  That sounds about right to me.  

Of course, we also still have easy access to the special-effect-on-doubles mechanic.  It's the degrees of success which get a little funky: something along the lines of subtracting half the larger die roll from 7, to generate a number between 1 and 5 (no 6 - if neither of your dice are higher than 1, you obviously didn't succeed - snake eyes is always a failure) with higher levels of success reserved to those for whom lower dice rolls can succeed.  (I could simply subtract the higher die roll from 13, but that generates a number between 1 and 11, but thats an awful lot of range.  Some skills use the degree of success for the number of questions the player gets to ask of the GM, for instance, and any more than 5 or so seems like it would bog down the pace of the game tremendously.)  Only a couple of skills, as I've written them, really make use of degrees of success, so this might not be a big issue anyway.

I'm still a little bit on the fence about this, so please weigh in: If you were running a game, which one would be easier, more fluid, more intuitive to use?  Is this the respect the humble d12 deserves, or is 2d12 for one of the game's core resolution systems just too weird to stick?  Would the moderate fiddliness of calculating degrees of success with 2d12 make you not want to use that particular mechanic?  Is there something else that strikes you as broken or unworkable?  Let me know in the comments!