Problem: Expertise breaks bounded accuracy. A lot. Like a whole lot. For example, it's easy to build a rogue with a Stealth modifier higher than most enemies'
passive Perceptions.
When there are challenges that some party members can't succeed at and others can't fail at, this makes it hard for the DM to build appropriate challenges for the whole party, and also can discourage players from participating. This is why I like bounded accuracy so much. (I know there are some people who feel otherwise, but screw 'em, this thread is about what
I would fix about 5E.)
Solution: Bounded accuracy works well in combat because, even though everybody's attack bonus is about the same, the
consequences of their attacks differ: different damage, range, damage types, special conditions, etc. But the outcome of skill checks is ill-defined; all we have is the modifier, so for a character to be better, they must have a higher modifier. Thus our solution must be to somehow introduce qualitative changes to skill checks.
Ability Check Criticals. When you roll an ability check, if the check succeeds, and you roll a natural 20, you get a critical result. If you are adding your proficiency to the check, and the check succeeds, then you get a critical result on a roll of 19-20.
A critical result gets the character the best possible outcome. It may grant bonuses above and beyond what the character was seeking. As a guideline, here are some things appropriate to a critical result.
- The check takes less than the normal time. For example, a lock that might take several rounds to pick only takes 1 round.
- The check affects more targets than expected. For example, intimidation aimed at only a single character winds up intimidating many.
- The check produces more materials than necessary. For example, instead of finding food for 5 people, the character finds food for 10 people.
- The character overcomes restrictions connected to the check. For example, if a check allows climbing at half speed, the character instead climbs at full speed.
- The character, or an ally, automatically succeeds at some future check. For example, if three successes are needed to research the location of a lost treasure, the skill check result counts as two of them.
- The character, or an ally, gains advantage on one or more future checks. For example, an attempt at deception is so successful that future attempts to sway those targets are at advantage.
Expertise. When you have Expertise in a skill (or tool), any success on a skill check is a critical result.
Why it Works: Skill and ability checks are open-ended, and defining them in clear terms like combat checks would require a LOT of text. Critical results are an equally open-ended alternative. Plus, D&D has needed degrees-of-success for a long time.
Allowing Expertise to grant critical results on any success seems powerful. BUT: It doesn't increase your
likelyhood of success at all. The rogue with Stealth Expertise is just as likely to succeed as the ranger with Stealth proficiency. Their failure is equally bad: they get spotted. But when the ranger succeeds, he's just hidden. When the rogue succeeds, maybe he hides as a free action; or maybe he can hide despite inadequate cover; or maybe he figures out a good hiding place, granting the rest of the party advantage on their hide attempts.