Do you want transparency on the fly? Play flat and expect that multiple TN mods will fuck up your system to where if you're fighting an invisible opponent on the deck of a rolling ship while under the effects of a poison you're not going to hit him unless you institute an auto miss/hit system... which neatly and as far as I can tell totally addresses all the BS about the superiority of RNGs and bell curves anyhow, considering the vanishing chances at the ends of the curve.
Do you want to be able to apply multiple mods with a reasonable belief that they do not automatically make any situation an impossible scenario? Use a bell curve. The GM probably won't know the probabilities off-hand and so you'll want to include a TN chart with qualitative and quantitative descriptions.
Here are some examples with an automiss/hit system:
baseline d20 mods with a base 11+ TN; {automiss, autohit} on {1, 20}
- invisible is a 50% miss chance = -5 to base, absent context
- rolling deck is a 20% miss chance = -2 to base, absent context
- nauseated is a 30% miss chance = -3 to base, absent context
= you only hit on a 20 (5% success rate)
baseline 3d6 mods with a base 11+ TN
- invisible is a 50% miss chance = -2 to base
- rolling deck is a 20% miss chance = -1 to base
- nauseated is a 30% miss chance = -1 to base
= you hit on a 15+ (9.3% success rate)
Seriously. That's the argument going on. That 9.3% on a sliding scale is better/worse than 5% on a flat scale and it's so obvious that anyone that is not pro/anti-skub is a fucking moron and should kill themselves and probably fucks children.
At values less than the RNG-fuckery, say a -9 to d20 (which still converts to 15+ on 3d6) you're actually comparing 10% to 9.3%.
At values greater than the RNG-fuckery, say a -12 to a d20 (converting to 16+ on 3d6) you're comparing 5% to 4.6%.
At values MUCH greater than the RNG, say a -14 on d20 (converting to 18+ on 3d6) you're comparing 5% to 0.5%.
Is the far-edge, vanishingly small portion of the NdX so much more awesome than dX's equally arbitrary automiss/hit values?
THE SKY IS FALLING.
...
It depends on context: you're not just adding bonuses, you're adding bonuses with an intended magnitude. And for bell curve systems where magnitudes are static and based on an even-chance baseline, you actually CAN add more static bonuses to a check despite having less discrete points. But you know that, since...PhoneLobster wrote:If your goal REALLY IS the ability to add bonuses to the roll multiple times it is simply NOT honest to say that more potential to add cumulative bonuses exists on 3d6, its just plain not true.
This statement here is the entire crux of the bloated discussion. The flat system adds flat numbers regardless of context and this means a +X is a +cX% every time. The curved system adds flat numbers but the +X becomes some +f(X)% dependent on context.More to the point why do you require the 1d20 example to add a +45% bonus each time when the 3d6 example is not required to do the same?
And that's the point.
Always adding +cX% means you run off the RNG quicker unless you're willing to recalculate the +X based on context, and at that point you're adding complexity on the backend of a flat system rather than the frontend with a curved system.
I have no idea why the math discussion is going in circles - everyone involved here understands it: PhoneLobster, Frank, Murtak, everyone. It's just that no one seems to get tired of telling each other that they don't understand it.