Can I salvage this core/dice mechanic?

[Greywander]

I've been thinking about dice mechanics and different ways to implement them. For example, D&D's d20 mechanic is elegant in its simplicity, being extremely easy to use, though it still has its own flaws. One aspect of this is that D&D's core d20 mechanic doesn't really offer any choices for how you roll the dice; all of the tactical decision making goes into deciding what actions you want to do, and once you've decided you simply roll to see if you succeed. I thought it would be interesting to explore giving players a level of control over how to roll, but the mechanic I've developed seems to have some issues.

Here's what I've got so far.

When a player performs an action (or otherwise needs to invoke the core resolution mechanic), they can choose to roll one, two, or three dice. If you roll more than one die, you use the highest roll and drop the other dice. As such, rolling more dice increases your chances of success. However, rolling doubles or triples will trigger a "setback" (more on these below). As such rolling two dice is considered to be "standard", while rolling one die is "cautious" (less likely to succeed but can't trigger setbacks) and rolling three dice is "reckless" (more likely to both succeed and trigger setbacks).

Attributes add their value to the number rolled (similar to D&D).

Your skill rank determines what type of dice you use. The lowest skill is d4, the highest is d12, incrementing one step each rank. Note that as your skill increases, not only does your average roll increase but you're also less likely to roll doubles or triples and trigger setbacks.

Rolling doubles triggers a minor setback. These tend to be inconvenient but not directly harmful. An example might be, "You swing your sword... and let go. It's now 20 feet away sticking out of a tree." Out of combat, these might waste time or trigger a random encounter ("You hear a patrol coming as you struggle to pick the lock").

Rolling triples triggers a major setback. These tend to be directly harmful. These range from damaging yourself or an ally to losing an item (the item breaks or is literally lost) or other unpleasantness.

You can both succeed on a check and trigger a setback at the same time.

Now let's talk about the problems with this system. One is that variance increases as your skill increases, i.e. higher skill = more randomness. Generally, higher skill should mean less randomness, but this is necessary to reduce the chances of rolling doubles or triples. I could invert it, using a roll-under system instead of a roll-over system, and have the die size decrease as your skill increases. This would make it less random, but it would ruin the setback system.

Another problem is that the grain of possible outcomes is too coarse. In D&D, a d20 roll can generate 20 different values (obviously), but with my proposes system you could only generate 12 values at best and 4 values at worst. A PC with a skill of d4 (and no attribute boost) can't beat a DC of 5, but if we drop the DC to 4 then they have a 25% chance of beating it (when rolling cautiously, it gets even worse if you roll more than one die). One possible solution to this would be to combine this system with a d20 roll, e.g. if you have a skill of d4 then you would roll 1d20+(1/2/3)d4h1, while if you had a skill of d12 then you would roll 1d20+(1/2/3)d12h1. But this makes the variance problem even worse, and it makes randomness (the d20) play a much greater factor than skill (the d4 or d12).

Is there any way I can save this mechanic, or should I go back to the drawing board?

[AtlasSniperman]

You could invert the whole system theoretically.
i.e. lower rolls are better. Roll multiple and drop highest etc. Keep the "doubles/triples are dangerous" thing, that works pretty nicely

and of course; the higher your skill, the lower the die type. d20 is worst, d4 is best. 1 is best roll. This will result in more skilled characters rolling cautiously(as the die size decreases, roll fewer dice to avoid rolling doubles) or allows you to open up ability options like "1/day ignore 1 die after rolling and before doubles and triples are checked"

And of course, in terms of the "DC" system; just invert how passing a DC is determined. DC 5 is hard as you have to roll [i] lower [i] than 6, and dc 15 is pretty much guarenteed for anyone using anything but d20's

Would that work?

Your idea sounds cool and I'd very much like to help iron out any quirks.

[Composer99]

To be honest, I think what you have identified as problems - variance and coarseness - aren't as problematic as you might think.

Part of the issue, I think, is that you haven't actually defined your set of typical DCs/target numbers, at least not in your post here. That would go a long way to determining whether these really are problems in the first place, or, at least, whether they are intractable problems requiring a bottom-up revision.

In d20 system D&D, especially 5e, most DCs are within the 10-20 range, and I hazard to guess that a majority of those are within the 10-15 range, so there is, in effect, a 6- to 11-value spread of reasonable DCs. There are, of course, some higher and lower outliers. If you define your spread of target numbers to be most often within the 5-15 range, then that goes some way to eliminating the variance problem: each increase in die size increases your ceiling of achievable results without hurting your floor. If most tasks of simple to moderate difficulty end up within the range of 5-10 for DCs/target numbers, and if you allow some kind of "passive" check feature in non-stressful circumstances, I think you'd be okay.

I daresay that defining the range of typical DCs/target numbers as being within the 5-15 range also addresses coarseness. After all, a typical range of 10-20 for DCs in d20 is pretty coarse: there are usually only 11 outcomes (and GMs are usually encouraged to use only three of those outcomes - 10, 15, 20 - for improvised DCs).

Having the typical DC/target number range being within 5-15 also makes it very clear how valuable advancing your skills can be: unlike in d20 systems, where you can always get lucky and roll high, in this system, skill really does matter. Say your attribute cap is 5: to hit a DC/TN of 15, you have to be able to roll a d10 or better for your skills: you can't hope to be a lucky amateur.

What I would recommend, before you scrap this system, is:
- Define your desired range of typical DCs/TNs
- Define your desired range of typical attribute modifiers
- Decide whether you get to have some sort of "passive check" (à la passive checks in 5e or taking 10/20 in 3.X) in routine circumstances

Once that's done, then you actually know the probabilities of hitting certain DCs/TNs with your various skill dice, and how that incentivises (or doesn't) taking gambles on doubles or triples.

[lightningcat]

Do you have any sort of random benefit mechanic? Something that provides the character some unexpected good luck, regardless of success.
"You are picking the lock to get into the room and hear the approaching guards strike up a conversation with one of the maids." sorta thing.

[Gorum]

I've been thinking about dice mechanics and different ways to implement them. For example, D&D's d20 mechanic is elegant in its simplicity, being extremely easy to use, though it still has its own flaws. One aspect of this is that D&D's core d20 mechanic doesn't really offer any choices for how you roll the dice; all of the tactical decision making goes into deciding what actions you want to do, and once you've decided you simply roll to see if you succeed. I thought it would be interesting to explore giving players a level of control over how to roll, but the mechanic I've developed seems to have some issues.

Here's what I've got so far.

When a player performs an action (or otherwise needs to invoke the core resolution mechanic), they can choose to roll one, two, or three dice. If you roll more than one die, you use the highest roll and drop the other dice. As such, rolling more dice increases your chances of success. However, rolling doubles or triples will trigger a "setback" (more on these below). As such rolling two dice is considered to be "standard", while rolling one die is "cautious" (less likely to succeed but can't trigger setbacks) and rolling three dice is "reckless" (more likely to both succeed and trigger setbacks).

Attributes add their value to the number rolled (similar to D&D).

Your skill rank determines what type of dice you use. The lowest skill is d4, the highest is d12, incrementing one step each rank. Note that as your skill increases, not only does your average roll increase but you're also less likely to roll doubles or triples and trigger setbacks.

Rolling doubles triggers a minor setback. These tend to be inconvenient but not directly harmful. An example might be, "You swing your sword... and let go. It's now 20 feet away sticking out of a tree." Out of combat, these might waste time or trigger a random encounter ("You hear a patrol coming as you struggle to pick the lock").

Rolling triples triggers a major setback. These tend to be directly harmful. These range from damaging yourself or an ally to losing an item (the item breaks or is literally lost) or other unpleasantness.

You can both succeed on a check and trigger a setback at the same time.

Now let's talk about the problems with this system. One is that variance increases as your skill increases, i.e. higher skill = more randomness. Generally, higher skill should mean less randomness, but this is necessary to reduce the chances of rolling doubles or triples. I could invert it, using a roll-under system instead of a roll-over system, and have the die size decrease as your skill increases. This would make it less random, but it would ruin the setback system.

Another problem is that the grain of possible outcomes is too coarse. In D&D, a d20 roll can generate 20 different values (obviously), but with my proposes system you could only generate 12 values at best and 4 values at worst. A PC with a skill of d4 (and no attribute boost) can't beat a DC of 5, but if we drop the DC to 4 then they have a 25% chance of beating it (when rolling cautiously, it gets even worse if you roll more than one die). One possible solution to this would be to combine this system with a d20 roll, e.g. if you have a skill of d4 then you would roll 1d20+(1/2/3)d4h1, while if you had a skill of d12 then you would roll 1d20+(1/2/3)d12h1. But this makes the variance problem even worse, and it makes randomness (the d20) play a much greater factor than skill (the d4 or d12).

Is there any way I can save this mechanic, or should I go back to the drawing board?

1. Your idea is not bad at all, but presented poorly. Contrasts are extremely important. When you're going to present a rather complex dice system where both the number and size of the die vary, don't talk about d20 being simple. d20 have unskilled characters beat skilled characters because of the absence of ANY form of bellcurve, binary results and a range of 20 different results being enormous.



2. If you work with a system of independent stats & skill (as compared to class and level), you can have the skill be matched with the dices (bigger die allowing riskier rolls) while having the stat add to the result. This way, if the target number is 5, a character with no skill but a powerful stat might roll his one d2 and add his modifier of +3 and have a 50% chance of succeeding. An highly trained character with no natural aptitude might roll 3d6+1 taking the highest rolled die, but risking a setback.

You might want to invert skill and natural aptitude tho. Someone skilled don't need to be as reckless as someone relying on natural talent.


3. Look into World of Darkness, New World of Darkness and Dream Pod 9's Silhouette system. All of these have dice pools that vary according to skill / stats. The first two had both the skill and stat (minus circumstancial penalties) rolled as one dice pool. Each die that rolled 7 - 8 or more scored one "success". In one of them (where the target number is 7), 1s were "botches" and retracted successes.

The Silhouette system used Margins of Success a lot. You rolled [Skill level]d6 + stat vs. your opponent's [Defensive skill]+stat. On a hit, your MoS was multiplied by your damage value then compared to your opponent's damage thresholds.

For example, a pistol might deal *15. My Stamina set my Light wound threshold to 10, my deep wound threshold to 20 and my instant death threshold to 40. Thus an attack with a MoS of 1 would wound me lightly and a MoS of 3 would kill me outright.

Now, if I wore armor, every threshold would be increased by a flat amount. Let's say 10. Now my thresholds are 20 / 30 / 50. Now MoS 1 hurt but don't wound me, MoS 2 skip the light wound and directly cause a deep wound, and MoS 3 don't kill me instantly as it is still in the deep wound threshold.

In that system, you incurred a debilitating penalty [edit: Read Setback, Glitch or Fumble] only if ALL dice showed 1s. This means that high skilled characters (with 3 to 5 die) would see such a dramatic event less than 1% of the time while dabbling characters ([2 dice, take lowest] @ skill 0 or 1 die) would face it 1/6th of the time or more. This is a much easier system to grasp, but lacks your idea of being able to succeed AND suffer a setback.


So yeah, I think it is salvageable, but you need to keep ALL the rest as simple as possible.

[continuumg]

There's certainly a solid foundation here. I'm sorely tempted to steal the main mechanic for a one-off adventure.