Apologies - I am still stuck up with Bounded Accuracy and Expertise. As far as I understand from the answers about what it means to break bounded accuracy, from the related Q&As about how bonuses to attack and skills scale, and what the design objectives for this were, a core idea of bounded accuracy is to keep challenges relevant and interesting throughout the PCs career:

  • Goblins remain a threat, even at higher levels, you do not need to change their stats

  • You do not need to scale DCs. A hard skill challenge remains hard, even at higher levels.

  • It avoids impossible tasks and automatic successes, so rolling the dice remains meaningful

  • It allows everyone to participate. The wizard may not deal quite as much damage as the fighter when resorting to their crossbow or cantrip, but they still matter and can meaningfully contribute. The fighter may not have as high a perception as the ranger, but they still have a chance.

This is achieved by the rather flat increase of the bonuses for both skill checks and attacks over the PC's career: barring magic items, they can be expected to increase only by 6 points, from +5 at level one, to +11 at level 20.

For combat, this seems to work pretty well. On the chart below, the blue line is the average result on an attack or check for a proficient character with a maximized ability score, and the black line is the expected AC of monsters as given in the monster building rules in the DMG (p. 274). You can see how the ratio remains about even for all levels, with the expected to hit about 2-3 points higher, resuling in an about 65% hit chance throughout. Instead of changing the hit chance, the game scales the hit points and damage, so a single hit from a mighty hero will kill a weaker monster. This works (mostly, PCs can find ways to build characters with very high ACs).

AC/DC vs Level and Skill Bonus Chart

For unopposed skill checks, the DCs are entirely static. Hard is hard is hard. The DMG disusses the matter as follows (page 238):

If the only DCs you ever use are 10, 15, and 20, your game will run just fine. Keep in mind that a character with a 10 in the associated ability and no proficiency will succeed at an easy task around 50 percent of the time. A moderate task requires a higher score or proficiency for success, whereas a hard task typically requires both. A big dose of luck with the d20 also doesn't hurt.
If you find yourself thinking, "This task is especially hard," you can use a higher DC, but do so with caution and consider the level of the characters. A DC 25 task is very hard for low-level characters to accomplish, but it becomes more reasonable after 10th level or so. A DC 30 check is nearly impossible for most low-level characters. A 20th-level character with proficiency and a relevant ability score of 20 still needs a 19 or 20 on the die roll to succeed at a task of this difficulty.

You can see from the yellow graphs for Good Stat or Proficient (average stat), that the DMG is right on this. These characters will have about a coin toss of a chance throughout their career with an moderate task ("medium" in the chart), a bit less early on, a bit more later on.

Proficient characters with good abilitiy scores will outgrow moderate over the course of their career, and be a match for hard challenges towards the end. One can expect that mighty heroes will confront more powerful enemies, with likewise better skills, so opposed checks might still scale. And these opponents may afford better traps or secret doors, so their lairs may sport hard challenges where normal dungeons had medium ones after about level 10, as the DMG suggests. That shift pretty much equalizes progress, and keep things interesting. So far, so good.

What I do not understand is how Expertise fits with this (shown as the red line in the graph). In particular in combination with Reliable Talent for rogues, a minium result of 22 or so after level 11 is guaranteeing automatic success on hard checks, on tier four even on very hard checks.

To me, this is clearly breaking the design objectives of Bounded Accuracy.

  • Automatic success invalidates die rolling

  • A hard skill challenge does not remain hard to solve at higher levels

  • Allowing everone to meaningful particpate breaks -- challenges for the rogue are impossible for others, challenges for others are a joke for the rogue

In my experience, what makes this even more problematic are cantrips like guidance -- in many cases, you can assume the roll will have 1d4 added to the result. With such boosters, even nearly impossible DCs can become near automatic successes1.

So, I am stumped. How does Expertise make any sense, given the objectives of bounded accuracy?

P.S. For my games, I have resigned myself to accept the fact that Expertise is effectively like magic. These characters just will detect all the secret doors, and will never be heard or seen when sneaking. That is fine. Magic can do similar things, it is not breaking the game.

P.S.S I apologize this came across a bit ranty. After reading up more on how to pose good designer intent questions, I realize it might have been better to frame it as a question on how to balance Expertise. I'll leave it stand as is, because changing it now would invalidate Ryan C. Thomas's excellent answer.

1 The same applies to magic items. In my experience groups tend to reinforce characters when they distribute magic items. So a skill enhancement item (like gloves of thievery, boots of elvenkind, eyes of the eagle etc.) will likely go to the expertise character. These items by themselves contribute to breaking bounded accuracy -- they grant +5 or even +10 bonuses, way more than the at best +3 bonuses from magic weapons. However, I think Bounded Accuracy assumes a game where magic items are optional, so maybe it is more appropriate to look at this without them.

  • 5
    \$\begingroup\$ I'm not sure if there's actual problem to solve, and it borderlines on designer's intent questions. I'm not the one who downvoted, but I kinda understand where it came from. \$\endgroup\$
    – Mołot
    Commented Jul 31, 2022 at 17:37
  • 2
    \$\begingroup\$ I did, and I did a VtC. As I see it, this discussion prompt would be better off at a forum, in my opinion. Given that expertise was originally confined to Bards and Rogues, and there was no feat that offered it, and that a few classes (Knoeledge Cleric and Ranger) had features that included something like it, the vast majority of PCs did not have access to it. I am not sure if it's a designer reasons at this point since later splats expanded it's achievability. \$\endgroup\$ Commented Jul 31, 2022 at 17:40
  • 7
    \$\begingroup\$ Justify it to who? What does it mean to make sense of it? You say it isn’t breaking the game, so I’m not understanding what problem we’re supposed to be solving here. \$\endgroup\$ Commented Jul 31, 2022 at 18:48
  • 5
    \$\begingroup\$ I don't see how you're getting "auto-success" for DC 30. I believe the max bonus from expertise plus stat is +17 (+5+6+6), which means a natural 13 is required for success on a DC 30 task. That's a 40% chance of success. This is still true even with Reliable Talent, because a natural 10 is a failure. It's true that an additional +3 bonus on the check from any source turns it into an auto-success with RT, though. Is that what you mean? \$\endgroup\$ Commented Jul 31, 2022 at 19:13
  • 3
    \$\begingroup\$ It's good to know designer reasons are on topic now, but this Q sounded kinda like a rant, without actual problem to solve, type of Q that made them off topic before. \$\endgroup\$
    – Mołot
    Commented Aug 1, 2022 at 20:02

1 Answer 1


Many abilities break bounded accuracy within a limited scope

Expertise is one of a number of abilities that "break" bounded accuracy, at least to some degree. Other notable examples include items like gloves of thievery (as you note), the Shield spell, and one of the most extreme instances, the cleric ability War God's Blessing:

At 6th level, when a creature within 30 feet of you makes an attack roll, you can use your reaction to grant that creature a +10 bonus to the roll, using your Channel Divinity. You make this choice after you see the roll, but before the DM says whether the attack hits or misses.

Not only do you get to add +10 to an attack roll, you get to do it after seeing the roll, so you can pretty much guarantee that every use of this ability will turn a miss into a hit. That clearly breaks bounded accuracy, but that is the point. The purpose of this ability is to turn almost any miss into a hit. In this case, there is an obvious limiting factor that prevents this ability from becoming unbalanced: it can only be used 3 times per rest.

Beyond large numerical bonuses, there are also some abilities that break bounded accuracy by literally guaranteeing success, such as the 20th level rogue feature Stroke of Luck (emphasis added):

At 20th level, you have an uncanny knack for succeeding when you need to. If your attack misses a target within range, you can turn the miss into a hit. Alternatively, if you fail an ability check, you can treat the d20 roll as a 20.

Once you use this feature, you can’t use it again until you finish a short or long rest.

Once again, this feature is balanced by having limited number of uses. (But that's cold comfort to the unfortunate surprised victim of the master assassin's auto-hit auto-crit sneak attack.)

Overall, the point here is that breaking bounded accuracy is OK if the scope of that breakage is limited in some way. Just like it's OK that an archmage can call down a meteor shower on the battlefield, as long as they can only do it once a day.

Of course, unlike the above examples, Expertise and some other features don't have a limited number of uses. However, their uses are limited in other ways. A rogue gets expertise in a only limited number of skills. If the situation doesn't call for any of those skills, their expertise does not help in that situation, and they are back in the range of normal bounded accuracy. Perhaps there's an argument to be had about how many skills a rogue should get expertise in, but again, when we compare against mages with meteors and wishes and such, expertise in a few skills doesn't seem so extreme. After all, being extremely skilled is one of the core aspects of the rogue class:

Rogues devote as much effort to mastering the use of a variety of skills as they do to perfecting their combat abilities, giving them a broad expertise that few other characters can match. Many rogues focus on stealth and deception, while others refine the skills that help them in a dungeon environment, such as climbing, finding and disarming traps, and opening locks.

Expertise isn't the only rogue ability that "breaks" bounded accuracy

In fact, even with expertise in 4 skills, there's another similar rogue ability that probably comes up even more often in practice: Evasion.

Beginning at 7th level, you can nimbly dodge out of the way of certain area effects, such as an ancient red dragon’s fiery breath or an ice storm spell. When you are subjected to an effect that allows you to make a Dexterity saving throw to take only half damage, you instead take no damage if you succeed on the saving throw, and only half damage if you fail.

Again, this feature is "limited" because it's only usable against damage sources that allow a dexterity save for half damage. However, such damage sources are quite common, and the ability is actually better than an automatic success, because it has a chance to negate the damage entirely, which isn't possible from an ordinary successful save. In other words, Evasion is like a better version of Expertise for a specific but rather common category of saving throws.

(Incidentally, if you want an even more extreme example, a 20th-level artificer can, with nothing but their own class features, stack a +27+1d4 bonus to their lockpicking check. +17 from abilities and tool expertise, +5 from Flash of Genius, another +5 from their infused gloves of thievery, and a d4 from Guidance. They have to roll a natural 1 and roll a 1 on the d4 in order to fail a DC30 check, a 1/80 chance.)

  • 1
    \$\begingroup\$ This is largely the answer I was going to write. I would have added that BA works best in broad cases like "attacking" or "spell saves" that will be used all the time in multiple circumstances by everyone. Where it breaks are, as you say, limited in scope; getting a Barbarian or Thief to fail a Dex save, for example. Let PC's be good at the few things they are good at. There are only four slots for Expertise and one of these is often taken for Thieves Tools. A wise DM will devise a broad range of challenges; when a Rogue has Expertise in Perception, some checks will call for Investigation \$\endgroup\$
    – Kirt
    Commented Jul 31, 2022 at 20:36
  • 8
    \$\begingroup\$ @SeriousBri I read it as "bounded accuracy is a base on which game is built, but every class gets some way to be more awesome than that in a specific way, and expertise is but one of these ways". \$\endgroup\$
    – Mołot
    Commented Jul 31, 2022 at 21:14
  • 9
    \$\begingroup\$ @SeriousBri More like "Bounded accuracy is a guiding principle and philosophy, not a dogmatic or religious principle; as evidence, consider these multiple other exceptions that are also part of the rules." \$\endgroup\$
    – Novak
    Commented Jul 31, 2022 at 21:31
  • 3
    \$\begingroup\$ Also, I was wrestling with whether or not to VTC this, even despite the new trial on "designer reasons" questions. This answer makes me glad I did not. \$\endgroup\$
    – Novak
    Commented Jul 31, 2022 at 21:42
  • 3
    \$\begingroup\$ @Novak QFT Bounded accuracy is a guiding principle and philosophy, not a dogmatic or religious principle; as evidence, consider these multiple other exceptions that are also part of the rules \$\endgroup\$ Commented Jul 31, 2022 at 21:55

Not the answer you're looking for? Browse other questions tagged .