The skill system in my homebrew system has gone through a few (more like many) revisions. I originally started with the skill list and mechanic from Worlds Without Number before paring it down considerable and then fleshing it back out.
Paring down the skill list seemed really appealing. Characters would have the minimal skills needed to be adventurers, and the remainder of their skills would come from an open-ended list where you could buy “skill specialities” for a narrow area. You might have (basic) survival skills, but you would need the speciality to track or do other non-basic things. This ended up being too confusing conceptually. Even I would mess up which skill to use from time to time.
I currently use a list of twenty two skills. I need to do a review of the social skills, so that may change a little bit, but I feel pretty good about the final skill list. Weapon skills are now also separate, and even armor is integrated as proficiencies (effectively, but they cost less EXP to buy). Unlike the previous way of doing things, this list is fixed. Specialities still exist, but they serve a different roll in the game. The hope is this will make determining the right skill easier to use for everyone.
Skill Checks are made using skill + attribute. The attribute is the approach you’re using. Are you using force (Strength), agility (Dexterity), smarts (Intellect), past experiences (Wisdom), fortitude either physical (Endurance) or mental (Willpower)? That’s the attribute. The skill is the method you’re using. Forcing open a chest with a prybar would be Burglary + Strength while fiddling with the lock might be Burglary + Dexterity or even Burglary + Wisdom with the right experience (“These are standard locks used on chests like this, which I’ve seen before.”). The choice of approach and method is left to the players, though they should change one or both if the table feels a suggested method + approach is not appropriate for the situation.
Wisdom is a bit of a special attribute because rather than represent something like intuition, it is your actual wisdom from past experiences. You get a fixed number of these at character creation (your background is one, then you decide two more). More can be obtained via longterm projects. Aside from working with the Wisdom attribute this way, experiences are also used for a handful of skills (Crafting, Performance, Rituals) to determine how those skills can be used.
I should also note that there is no “Perception” skill. Things like sneaking or surprise work through other mechanisms. I don’t really care for the loop where the stuff to do in the room is hidden behind a skill check I request as soon as they enter the room. I don’t think take 10 or passive checks are a good solution for this. I’d rather just describe the situation and let the PCs interact with it. When they want to learn more, they can use the Investigation (encounter) and Research (long term) skills.
Dices mechanics have varied considerably. As noted above, I started out using 2d6 from Worlds Without Number. The skill system in those games is inspired by Traveller. I wanted to use a fixed target difficulty and eventually degrees of success, so I have tried several different rolling methods to accommodate the range of modifiers in my game (2d6 → 3d6 → 2d6 → 2d10 → ??). That question mark is where I am at currently.
One problem I have encountered with degrees of success is the perception by one of my players that “mixed success” means he failed. Even though his character does what he wanted, the fact that something else happens feels bad. To address this dissonance, I’m considering a mechanic I am calling “dual rolls” for our next session. With dual rolls, instead of his rolling alone, I would simultaneously make a roll for the consequence. Based on the combination of rolls, this determines the result.
The idea is decoupling the consequence should make it feel better to players like the one in my group. As part of the system’s design, I don’t want to decide what happens. I’m only supposed to do that when the system requests it (part of my attempting to be a neutral referee by keeping my hands off the till). The consequence is my space for doing that, and for doing that as hard as the situation merits. It should be thought of as something akin to a wandering monsters check except systemic.
I’ve done some looking at the math and done one mock combat so far. I’d like to also mock out some of the conflicts I’ve described in the commentary thread (particularly
this one and
this one). The combat went okay. It’s a little more dangerous because the target rolls one of dodge, block, or parry. I’m currently using 1d20+mods versus 11 or 21 (for a Critical Success). This means success is “easy”, but it’s not always going to be consequence-free.
In the mock combat, Dingo took a bit more damage than he did in the actual session (due to some bad dice luck where both he and Deirdre whiffed). One thing I did like is it makes it possible for even those with crappy values to have some success attacking and defending. Trying to balance a progression of modifiers without instituting a difficulty threadmill has been challenging. Dual rolls obviates that. You can do some fiddling with the opponent’s roll (Deirdre’s Unbalance speciality was applied to the target’s next dodge), but they’ll typically get to make their roll.
As an aside, I’m not aware of many games that use this approach.
@pemerton has brought Pendragon’s opposed rolls to my attention, which has been the closest. Most games that use opposed rolls compare the results. My approach does not. That’s how it can generate the different degrees of success and how consequences are still possible even when you’re very likely to get what you want. Otherwise, the system breaks down when actions never have consequences (and needing to determine the consequence roll becomes an easy and visible test of whether anything is at stake).
Update: After doing some more testing and consultation with the player who had issues with the 2d10 method, we’re actually going to stick with the current method. The “dual rolls” method has some edge cases relating to things like spells that would result in quite a bit of upheaval in the way I’ve designed specialities and spells. I can do some things to better signal what consequences are coming, and I need to better tune the monster numbers. The rest of this post should still be applicable.