Dilemmas...

[Georgedoe]

Can anyone offer any advice/insight with this problem?

[Onix]

So the main problem is the jump in between skill levels.

Another possibility that would be to allow for decimal skill values (over 1 of course). That makes for ugly math that most players won't want to deal with unless they're all equipped with a calculator.

A better solution is to emulate the decimals with pips like in Star Wars (second ed at least I haven't played newer eds). Unfortunately the pips would have to be all different values for each skill level but you only have 5 so a chart with the pips wouldn't be out of the question. For skill level 2 it could have 2 +5 for the skill level. So you roll 3D10 * 2 +5. I haven't done a math analysis on that but I can if you want, Excel is what I use for massive dice rolling. Maybe the first pip is +3 then +5 for skill level 2 (or +2 and +4, whatever makes sense).

If that's unclear let me know.

[Georgedoe]

I would rather not use pips and charts per se, but your idea just gave me another possible option.

You roll a main die, which is the one that gets multiplied by the skill rating, and all your other dice serve as auxiliaries. The auxiliaries may add let's say +2 to your main die if they score 7+. This is a hybrid success system.

For example, if you make a 3Dx4 roll, you would roll 1 main and 2 auxiliaries. If you score 8 on the main die and 9 and 5 on the auxiliaries, your final result would be 8+2=10 x4=40.

The use of a single die to determine the main result breaks up the consistent distribution of numbers when you pick the highest out of 3 or 4 dice, which was the main problem of my system. A higher multiplier will still give you a good edge, just not as big as before.

[Onix]

Not bad, but the character with the higher skill level would also get this boost. This would have an effect on a contest between a 3Dx4 vs 4Dx3. If thats what you're after, then I think you hit your mark. But I don't think it will significantly effect a 3Dx3 vs 3Dx2. The likelihood of the lower skill not beating the higher skill is going to remain unchanged from things as they are.

[Georgedoe]

Not quite unchanged, since the main die is only one die. You might score 1 or 2 and then you'd be screwed even with a big multiplier.

I've run the main scenario on AnyDice using "output d10*2 - d10*3" and it gives a 30% chance for a skill 2 character to beat a skill 3 character, which I think is reasonable (it's a huge improvement over the 12% I got using the old system).

[Onix]

Thirty percent sounds like the target you were aiming for and it sounds reasonable to me. Hmm, it's interesting that it changed the probability, I'll have to play with that just to satisfy my own curiosity. Surprising probabilities are really cool because they can model all sorts of different things.

[Georgedoe]

I spent most of the morning refreshing my Qbasic skills and writing a small program (small but took longer than I thought) to simulate the following scenario:

Roll a d10 and multiply it by the skill rating. Roll a number of auxiliary d10s equal to your attribute. 7+ on an auxiliary die add +1 to the multiplier, 10s are rerolled.

The program simulates 100000 resisted rolls.

So far I got:

3Dx2 versus 3Dx3 has a 36% vs 59% chance to win, the rest being ties.

3Dx2 versus 3Dx4 has 28% vs 68%.

4Dx2 vs 3Dx3 has 41% vs 54%.

4Dx2 vs 2Dx4 has 36% vs 60% (notice that the 2-point skill advantage is still superior to the 2-point attribute advantage).

And finally, the extreme hypothesis of 5Dx1 vs 1Dx5, superb natural talent versus excellent training and experience, which wields 26% vs 70%.

I think this is by far the best results I got that still fall within the main theme of my system (that is, skills giving a greater specific advantage than attributes).

[Onix]

So you wouldn't choose the die you multiply anymore?

[Georgedoe]

No you wouldn't choose. This has the side effect of having a chance to fail even simple tasks if you roll low. I don't think that's a bad thing though. Success should not be taken for granted even with high ratings.