I've been messing about with a new idea for the past few weeks, so I'm going to start sharing some bits of it on here to see what you guys think.

First up, the basic game mechanic.  I'm aiming for something relatively quick and simple, and decided to go with a d% for the actual roll but wanted a way to use smaller numbers for the things like stats and skills that determine what the % chance of the check will be.

Here's a 3 page pdf with the basics explained:

http://drive.google.com/open?id=1rOMlLeiRkEWpOWqrVj6YRgxgCvBQgyFm

I'm not going to get into the setting or game concept yet. I want to run some mechanics and other ideas past people first while I flesh things out a bit more and make the writing neater.

One thing I noticed looking at the mechanics is that the slope of your curve is steeper once success chance drops below 50%.  If Level is 10 points higher than Target, for example, you improve your odds by 20%, but if Target is 10 points higher than Level your odds are reduced by 28%.  Is there a reason for this?

I was also curious as to what was considered the "average" ability score for a starting character.  If a reasonably competent professional has a total Level 12, for example, he's only going to succeed on a Moderate difficulty Target (4) 66% of the time, i.e. he's going to fail one out of three rolls at that level.  Even on a Very Easy (1) Target, he's still failing about 1 in 4 rolls, which is a 25% failure rate. To get the kind of reliable performance that people tend to show in real life at their trained areas of expertise -- i.e. considerably below a 10% failure rate -- it looks like you need a Level of at least 17 or 18 vs. Targets no higher than 3, and your rules imply that this would only happen very rarely since the limits of these scores are basically 1 to 20.

One way around this might be to say that for any situation where Level exceeds Target and the situation doesn't involve active stress, you can automatically take a number of successes equal to (Level - Target) without rolling -- this would be, effectively, your equivalent of the "Take 10" rule.

Another would be to reorganize your Level vs. Target dynamics not around a flat mathematical progression from the "Level = Target: 50%" baseline, but around what various Targets mean to a specific Level. In general, once a particular task becomes more easy than not for a particular level of ability, the average person's capacity to perform well at it tends to go up sharply and then plateau as skill increases.  Similarly, once a task exceeds our ability, our chance of success drops off markedly.  I'd suggest working out a progression by setting Level vs. Target thresholds for the following probabilities:

- Success Virtually Guaranteed (98%) -- this is the chance of rolling anything but a 3 on 3d6.
- Success Mostly Likely (83%) -- this is the chance of rolling anything but a 1 on 1d6.
- Success More Likely than Not (67%) -- this is the chance of rolling 3 or higher on 1d6.
- Success is a Tossup (50%) -- the chance of rolling 4 or higher on 1d6.
- Failure More Likely than Not (33%) -- the chance of rolling 5 or higher on 1d6.
- Failure Mostly Likely (17%) -- the chance of rolling 6 on 1d6.
- Failure Virtually Guaranteed (2%) -- the chance of rolling 18 on 3d6.

If you want "Level = Target" as your 50% baseline, then you could set a 10 point difference as what moves you from "Tossup" to "Virtually Guaranteed" either way: i.e. at "Level = (Target + 10) Or Higher", you have 98%, and at "Level = Target - 10 (Or Lower)" you have 2%.  Then set the points where you want the curve to break and distribute the per-level steps evenly between them. For example:

- For an even progression, set "Mostly Likely" (83%) at "Level = Target + 6" and "More Likely than Not" (67%) at "Level = Target + 3".  Each step up or down the Target scale is going to increase or decrease your odds by about 5%.

- Or, for a steeper curve, set "Mostly Likely" (83%) at "Level = Target + 5" and "More Likely than Not" (67%) at "Level = Target + 2".  From "Level = Target + 10" to "Level = Target + 5", each step will reduce your chance by about 3%, but from Target + 5 to Target + 2, each step reduces odds by about 5%, whereas from "Target + 2" to "Target" each step reduces your chance by nearly 10%.

- For a really steep curve, set "Virtually Guaranteed" (98%) at "Level = Target + 6", "Mostly Likely" (83%) at "Level = Target + 3", and "More Likely than Not" (67%) at "Level = Target + 1" (!). This makes the determination of whether you're within one or two levels of the Target really important, and can make looking for that last vital +1 a real issue.

When you begin your explanation, include a short glossary of terms. This way, you don't have to be so esoteric in the opening paragraphs.

d100 / d% - use two ten sided dice, designating one as the tens, ... yada yada.

Then, you can remove the "check" verbiage and just go with a Task, where you roll d% and reference the table.  You use the term "roll a dice" which should be roll the dice but really should be "make a Task roll" or something easier.

This table looks a lot like the BRP resistance table. You may want to peek at it for better math as Mr Tannhauser pointed out. From a basic level, I really dig the "compare two scores and make a check" idea. I think that's slick. I wouldn't change dice or change the "use a table" approach, but just tweak it to address potential wonkiness. Also, your levels of success idea is really cool. I feel like your onto something here. :-)

@Stephen Tannhauser, thanks for that! Exactly the kind of feedback I was hoping to get from someone with a better appreciation of the maths than I have. Your comments will definitely be taken into account as I revise the table and the numbers.

@trechriron, thank you also! Words of encouragement from a veteran like yourself are always appreciated!