Reverse Engineering (2025)

I've been thinking on how to handle the skill check mechanics for the sequel, and I realized the original idea had gotten away from me, somewhat.

In TNP, the skill "check" die is a d10, and the bonus die is a d6; to mimic this in a d6 system, the obvious thing to do would seem to be replacing the d10 with 2d6. But somehow I'd gotten to a place where it had morphed into some variation on "d6 pool for everything, all the time," and it still seemed to miss the mark. Without beating a dead horse, the effective mod range is a bit limited, and the dice rolls that matter are likewise pretty limited. So I went back to the fundamental 1d10+1d6 idea.

Ok, let's say the "check" die is 2d6; anytime you make a check, you roll 2d6. Against a DC10, this gives you only a 1/6 chance of success -- so some more dice and/or mods are clearly needed. The average on 3d6 is 10.5, meaning that 3d6 vs. a DC10 is basically a coin flip. What TNP does is essentially increase the number of "bonus" dice you can roll, thus increasing the odds of rolling a high bonus/modifier to your check. We could apply this same logic to a 2d6 system; the base check is 2d6, and the bonus is the highest d6 out of a pool. If we're using the current baselines, this would be a maximum of +2 from Attributes, and probably the same for Skillsets (maybe +3?)

Using anydice, we can figure out the odds for how this works vs. a DC10:

• 2d6+1d6 = 62.50%
• 2d6+[highest 1 of 2d6] = 75.23%
• 2d6+[highest 1 of 3d6] = 81.13%
• 2d6+[highest 1 of 4d6] = 84.37%
• 2d6+[highest 1 of 5d6] = 86.35%

This actually looks really promising, because as the chance of failure drops towards that 15% threshold, the diminishing returns on further boosting a skill start to kick in. Adding the implicit assumption of a 3rd die (except in the case of a +0 mod) also gets around the problem of "only 5s and 6s matter." The problem is that this breaks the check into 2 distinct rolls, which is a bit inelegant; the base 2d6 clearly can succeed, but is unlikely to do so -- but pooling them all together completely changes the math. If the 2d6 is a 3 or less, then the bonus dice become irrelevant (because the check has no chance of success) but a result between 4 and 8 means that success is still possible -- unlike with the "dice pool only" model, where it's just... 2 out of X dice have to equal 10, so if you roll all 4s and 3s (or lower) you're just screwed.

So now that we've established that this works, the immediate question is, can we further reverse-engineer this into a combat system? As I've often said, you should either have a unified system (that usually works great for one subsystem and badly for another) or you have distinct subsystems, but both serve their purposely excellently. The conundrum of an "all d6 system" is that it's not very unified if there are still distinct mechanics for combat vs. for skill checks -- despite everything using the "same" dice. 

The current combat mechanics are based around the supposition of 3d6+mod, where 2 dice (plus a flat mod) are used for the attack roll, with the remaining die (plus the same mod, probably) being used as the damage roll. To make this work anything like the check dice... basically it would mean a significant boost to hit-chance, assuming we're sticking with a cap of +2/+2d6 modifiers (which, you can't really meaningfully go lower than that, so...)

As mentioned in the previous post, the fact of attack rolls having damage as a "release valve" for unused dice means that the attack mechanics will likely/necessarily have to be different than the skill check mechanics. The other consideration is that the "acceptable ranges" for both types of rolls are different; skills should mostly fall in the 45% (i.e. a -1 mod, in D&D terms) to 85% range, whereas attacks probably need to have 55% or 60% as a baseline, and increase to 90% or possibly 95% with teamwork and other bonuses. This all makes it hard to unify mechanics, since unified mechanics should (presumably) produces unified outcome ranges -- so if that isn't the goal, then unified mechanics likewise shouldn't be the goal. This new idea now gives us mechanics that hit both of our prescribed benchmarks; the question next is can we do anything to further simplify and/or unify these two systems, and still produce comparable outcomes?

...

Next post is due on (or about) March 15th, so check back then for more!