Eventually, someone is going to try something that is not absolutely certain to fail or absolutely certain to succeed. It is, however, more or less certain to do one of these.
Existing Resolution Models
Here is a brief look at how some games handle task resolution.
In Dungeons & Dragons 3e, tasks were resolved by rolling d20, adding modifiers (possibly many modifiers), and comparing the result to a Difficulty Class (DC). Whether this was attacking an enemy with a weapon, climbing a wall, or resisting poison, they were all resolved the same way.
There were exceptions. Some checks made no allowance for character skill or natural ability, they were beyond the control of the character doing the action. The most common one was probably attacking a concealed creature. Partial concealment gave a 20% miss chance, and full concealment (you have a pretty good idea where the target is but cannot see the target) gave a 50% miss chance. No amount of skill could overcome that. These checks were made with percentile dice to help highlight their difference.
And there were exceptions to those exceptions. Certain abilities would let you ignore a particular chance of failure, such as detect invisibility letting you negate that ability and attack normally.
A d20-based resolution has linear frequency on the rolls. That is, there is a 5% chance of getting each of the values possible for you to roll. Pretty simple to calculate, but ‘very swingy’ and you need to get a fairly high modifier to have a reliable effect on the outcome. Even DC 11 (50% chance of success with no modifiers) requires a +5 modifier to have a 75% chance of success. Combat can have critical hits as well, but there was typically a confirmation roll needed so attacks with low chance of success didn’t have disproportionate critical hit frequency when successful.
That is, if you need a natural 20 to hit, and natural 20 is a critical hit, then you’re not going to hit very often at all, but every hit is a critical… confirmation checks mean you’d need to hit again to get the critical (1/400 chance). Someone who attacks with +9 will hit on 11+, half the time, and get a critical threat on 10% of all hits and confirm half of those: 1/20 of all attacks would be critical hits.
Advanced Dungeons & Dragons Resolution
AD&D was all over the place. Thief skills were roll-under percentage checks, attack rolls were roll-over d20 checks (subtracting the target AC from the character’s THAC0 and comparing the roll + modifiers to the result), proficiency checks were roll-under d20 checks against one of the character’s ability scores (possibly modified).
Any particular subsystem could use different resolution mechanics than the others. The d20-based games did have some exceptions to the basic rule, but they were for specific purpose and consistently applied.
Hero System Resolution
Hero System uses a 3d6 roll-under system, with target numbers based on the character’s ability score and modified by circumstance (typically Requires Skill Roll powers apply a -1 penalty for each 10 Active Cost).
When I was in college I started work on a variation that inverted the checks. Instead of targets being ‘9+score/5 or lower’ (STR 10 means you succeed on an 11- roll), you rolled 3d6+score/5 vs a target number of 12. Things that made it easier increased your roll, things that made it harder increased the target number. Your 40-point RSR Energy Blast would mean you roll 3d6+INT/5 (3d6+3 for my INT 13 wizard — 2.6 rounds up to +3) versus a target number of 16.
There were also ‘base value’ targets, such as a skill familiarity meaning you succeed on 8-. Your ability scores don’t come into it, it just means you know a little bit about the task and can handle simple things pretty reliably, or more complex things on a hit-and-miss basis.
Being a 3d6 system, modifiers have some greater effect, especially in the middle range. That is, if you need to roll 11+ on the dice to succeed, even a +1 means you now succeed on 10+ — a 12.5% improvement! It has much less effect if you are very unskilled (need a 17+ to succeed, +1 means you need 16+ to succeed, an improvement of about 2.8%) or very skilled (4+ to succeed, +1 means you now need 3+… yes, that means you cannot fail, but it’s a difference of only 0.46% to your success rate), and I’m quite okay with that.
Basic Roleplaying System
Percentile roll-under system. Moving on.
Green Ronin’s AGE System ends up being fairly similar to the Hero System variation I came up with in college, with a very interesting wrinkle.
It’s still 3d6+modifiers versus a target number, but one of the dice is a different color. On a successful check that contains doubles (any two of the three dice have the same value — 96/216 chance with 3d6, if we ignore the success requirement), the character gets a number of stunt points equal to the value of the different-colored die. These stunt points can be used to increase or modify the effect: on an attack can give bonus damage, or an extra attack, or knock the target prone, or let you move yourself or your opponent, and so on.
I really like this. The times I’ve played AGE, it gave a fluid mechanism for extra effects, without needing an explicit ‘critical hit’ mechanic. There were standard stunts available, but we treated them as examples and guidelines to what certain stunt point values should be able to do. When running a game I’d look at a player’s request and decide if it was a reasonable effect for the points.
Powered by the Apocalypse
In games powered by the apocalypse (Apocalypse World, Dungeon World, etc.) there is a very simple mechanism. Roll 2d6 and add modifiers (usually not very many, maybe up to three points worth).
- 10..12: you succeed
- 7..9: partial success: “you do it, but…”
- 2..6: failure
This is… okay. I like a little more symmetry in my results, and stumbled on (thanks to ‘related posts’) an Apocalypse World hack I wrote about eight years ago.
Still roll 2d6 and add (small) modifiers, but the results are as follows:
- 12+: great success: “you do it, and…”
- 10-11: success: “you do it”
- 7-9: partial success: “you do it, but…”
- 4-6: failure: “you don’t do it” or “you don’t do it, but…”
- 2-3: total failure: “you don’t do it, and…” or “you don’t do it”
This adds a little more texture the results, allows for better than expected results, and the possibility, slim as it might be, of making the situation even worse. This probably doesn’t come up much because you almost need to have a penalty for it to matter (1/12 chance to roll 2..3 on 2d6 without modifiers, 1/36 with a +1 and 0/36 with +2).
At this point, I’m leaning toward something like the AGE System. I really like the stunt system, and the bell curve in the rolls greatly reduces the ‘swinginess’ of the results. Even a +1 modifier has significant effect when things are most in doubt (i.e. close to 50% chance of failure) and I like that it has less effect when there is little doubt (so good it won’t matter, or so bad it won’t matter).
This means I’ll have to keep a close eye on modifiers. In a 3d6 model like this, I’ll want modifiers to be roughly half the size they would in a d20 model.
I could go back to a d20-based model. It’s familiar, of course, and easily applied. I could also explore options I didn’t even describe above, such as dice pools and the like, but I’ll save that option for later.
I devised the Advantage Dice System for Echelon, which is a dice pool that integrates stunt points. You start with a single die based on your tier (band of levels), and can gain additional dice for your talents, the abilities you have chosen. This increases your chance of success in the first place, and each successful die after the first gives you an increasing number of stunt points (1 stunt point for the first extra success, 2 more for the second, 3 more for third and last).
This is a bigger deviation than I’m prepared to make at this time, so I’m not pursuing it. For now.