D&D (2024) Should Bounded Accuracy apply to skill checks? Thoughts on an old Alexandrian article

[clearstream]

Running back for a couple quotes like this. My problem with this presentation is that it leaves out a really, really important element that (for example) Dungeon World makes explicit: uncertainty alone isn't enough. It needs to be uncertain and interesting. If you as DM simply cannot think of an interesting consequence for failure, don't roll. If you can't think of an interesting consequence for success, don't roll. Uncertainty is a necessary condition, but it's not sufficient.

100% agree, and on that the following 2024 PHB text is apposite

The DM and the rules often call for an ability check when a creature attempts something other than an attack that has a chance of meaningful failure. When the outcome is uncertain and narratively interesting, the dice determine the result.​

See, here's your problem. You've assumed, IMO quite wrongly, that DMs would never do that because it would be boring. The actual practice couldn't be further from the truth at many tables. Doesn't matter that the text doesn't oblige it. Tons of DMs do.

Local cultures of play can vary significantly. Here, I am focusing on what the rules demand.

And to be clear, it isn't just combat where this is a problem. Skills in 5e--which explicitly tell DMs NOT to run it like this!--are done in a similar fashion, run much more similarly to how they worked in 3e, despite having a skill system that is closest (not really THAT close, but certainly closest) to 4e's. Skills are treated as incredibly hyper-narrow things, the DCs are so frequently sky-high to do anything remotely useful or interesting, and the old scourges of things like iterative probability ("Roll for stealth....every single turn") and single-failure conditions ("if anyone fails this group check, the group fails") are back in full force. I consider myself profoundly lucky to have a 5e DM that uses reasonable DCs and takes a wide view on what skills are actually capable of.

To my reading, the 2024 text avoids encouraging hyper-narrow interpretations of skills or calls for checks that are uninteresting better than the 2014 text does. Frex, consequences resolution is shifted from the DMG into the PHB. Some posters are vexed to have skills less tightly defined: I read that same light touch as a conscious choice aimed at encouraging less hyper-narrow application.

Personally, I think the example is more than a little wrong due to stacking together many things that won't actually be available every game, let alone every check. You've also misconstrued Reliable Talent as +10 to +20, when it's actually +0 50% of the time (any roll higher than 9), and anywhere between +1 and +9 the remaining 50% (any roll from 9 down to 1.)

I intended my words to be understood as meaning that instead of the contribution from the d20 being +1 to +20, it becomes +10 to +20 (ignoring the distribution, as I did throughout.)

Bardic Inspiration depends on specifically having a Bard; Psi-Boosted Knack depends on you yourself being not just a Rogue but a specific kind of Rogue.

Agreed, but my point is to show what a party might well have. I take it you do not disagree that those mechanics are available to player characters and do factually represent the possible (regardless of how common) range? I'm running a short campaign right now to understand the 2024 rules. Guidance features prominently, but we have no bard, ranger or rogue so no one has expertise.

Kicking those out as being too situational (at least guidance is a cantrip several classes/subclasses can learn) and fixing your incorrect statement regarding reliable talent drops the top end by a whopping ~22 points, down from allegedly +52 to "merely" +30. Which, I admit, is still extremely high! But let us not pretend it is that ridiculous in anything but the rarest of cases. I may be a vociferous critic of 5e, but I'm not going to build an argument against it on something like this.

I suspect that you are taking my numbers to be the modifier: they're not, they're the rollable result. Hence the missing "+"... I literally intended (and in fact typed) "-4 to 52".

+1 to +20 for the d20 (which becomes +10 to +20 with Reliable Talent, so does not change the top end)​

+2 to +6 for proficiency​

+2 to +6 for expertise​

+1 to +12 for Bardic Inspiration or Psi-Bolstered Knack (I assume only one can apply)​

+1 to +4 for guidance​

-4 to +5 for ability score (player characters cannot have 1 or 2, ordinarily)​

​

The lowest number I can roll is +1 -4 = -3. Throwing in the kitchen sink yields up to 53. So the range of rollable result is -3 to 53. The range of modifiers is -4 to +33.

 

[clearstream]

Sorry, but this is hilarious! I didn't even know they moved it from 11th to 7th. Sigh... man, that is so wrong and another "power creep" bump! Yeah, 2024 D&D will NEVER see the light of day in my games....

Yup, it's seems like a conscious choice to make tier-2 rogues able to essentially knock most doors and spider climb most walls (without spending a resource.) If that is right, then it amounts to a choice about the style of play they wanted to offer rogues.

 

[ezo]

Yup, it's seems like a conscious choice to make tier-2 rogues able to essentially knock most doors and spider climb most walls (without spending a resource.) If that is right, then it amounts to a choice about the style of play they wanted to offer rogues.

Oh, I know. This "style of play" is rampant throughout all of the 2024 PHB, I just ( I am still laughing about this...) didn't realize this was another offense. Forget the concept that 2014 PCs can play alongside 2024 PCs...

Evasion is a fantastic feature for 7th level, one most Rogue and Monk players see as a real milestone. Adding Reliable Talent on top of that? Ridiculous IMO.

 

[EzekielRaiden]

I think this discussion is missing that bounded accuracy isn’t just applicable to the “skill system” in 5E—a term that I’m putting in quotes, as it’s pretty far from being a skill system in the sense of its use in D&D from AD&D OE where it was introduced through 3E. It is, at best, is a semi codification of the B/X system of “how to adjudicate things not in the rules.”

Bounded accuracy also applies to AC and to hit modifiers. It also includes the HP bloat of foes to account for the cap on AC. It also capped ability scores.

Looking at the effects of bounded accuracy solely within the frame of skills is really missing the forest for the trees.

Failing to look at the system holistically isn’t a really useful way to judge its performance.

I already consider the "progression in monsters is only through HP" to be a problem.

Considering it just in skills is me being kind.

 

[EzekielRaiden]

Local cultures of play can vary significantly. Here, I am focusing on what the rules demand.

I mean, the rules can demand all they want. If the majority of DMs don't do it that way, the text isn't really having any effect, is it? We already saw that problem explicitly called out by the 5.5e designers, e.g. Crawford openly stating, on camera, that people weren't taking enough short rests and so classes like Warlock and subclasses like Battle Master were getting short shrift, despite the rules actually saying you needed enough encounters and short rests to make things work.

To my reading, the 2024 text avoids encouraging hyper-narrow interpretations of skills or calls for checks that are uninteresting better than the 2014 text does. Frex, consequences resolution is shifted from the DMG into the PHB. Some posters are vexed to have skills less tightly defined: I read that same light touch as a conscious choice aimed at encouraging less hyper-narrow application.

I agree that the text intends this. I just think that intent has been completely ineffectual. Whether that's insufficiently talking about it, or talking about it but being soundly ignored more than half the time, or being overlooked, or whatever, I've no idea. Again: the rules can demand all they want, but if people aren't doing that with any degree of consistency, what does it matter what the rules demand?

I intended my words to be understood as meaning that instead of the contribution from the d20 being +1 to +20, it becomes +10 to +20 (ignoring the distribution, as I did throughout.)

Then I blame being tired and distracted, but see below.

Agreed, but my point is to show what a party might well have. I take it you do not disagree that those mechanics are available to player characters and do factually represent the possible (regardless of how common) range? I'm running a short campaign right now to understand the 2024 rules. Guidance features prominently, but we have no bard, ranger or rogue so no one has expertise.

Sure. I just think it's not really much of an argument to talk about the absolute bleeding-edge maximum; that ignoring the distribution is a pretty serious fault (since with this many dice and such, average results are much more likely than maximum results); and that looking at the spread when every WotC edition has always had a bare minimum of, at least in theory, -4 to +25 on rolls regardless.

I suspect that you are taking my numbers to be the modifier: they're not, they're the rollable result. Hence the missing "+"... I literally intended (and in fact typed) "-4 to 52".

+1 to +20 for the d20 (which becomes +10 to +20 with Reliable Talent, so does not change the top end)​

+2 to +6 for proficiency​

+2 to +6 for expertise​

+1 to +12 for Bardic Inspiration or Psi-Bolstered Knack (I assume only one can apply)​

+1 to +4 for guidance​

-4 to +5 for ability score (player characters cannot have 1 or 2, ordinarily)​

​

The lowest number I can roll is +1 -4 = -3. Throwing in the kitchen sink yields up to 53. So the range of rollable result is -3 to 53. The range of modifiers is -4 to +33.

Should we really be considering the die as "+1 to +20"? That's...kinda giving the single biggest contributor here. As in, almost half of the absolute maximum someone could get, after ultra-hard hyper-focus on skill bonuses and nothing else. The +33 number is much more useful for most people, since everyone has always been able to crit in D&D, as far as I'm aware.

Though I admit it is pretty funny that this means 5e has ended up closer to "4e times three fourths" rather than "4e cut in half or less", mathematically speaking. Kinda puts to question whether the real issue with 4e was the size of the numbers, or the amount of bonuses being added. The two things aren't the same, and even I recognize that, despite my love of crunch, it is possible to have too many fiddly math gewgaws. (Of course, I also believe the rampant over-use and abuse of Ad/Dis is another unwise design choice of 5e--a design choice I foresaw long before 5e was even published.)

 

[Parmandur]

Focusing on rogues specifically, they have Reliable Talent from 7th, which gives

Tier 1: +7 > DC 10
Tier 2: +11 > DC 24
Tier 3: +15 > DC 28
Tier 4: +17 > DC 30

I took that to be @Parmandur's point in their #37

Yeah, and the 30 cap is quite consistent across the game math for anything involving a d20 test. Both the 2014 and 2024 DMG do describe how the odds work for lower level to high level and proficient etc. characters up to 30. Having an Expert Rogue be able to hit 30 DC with some regularity is part of Bounded Accuracy, not breaking it. It is just...the upper bounds of accuracy.

 

[Parmandur]

Yup, it's seems like a conscious choice to make tier-2 rogues able to essentially knock most doors and spider climb most walls (without spending a resource.) If that is right, then it amounts to a choice about the style of play they wanted to offer rogues.

Yup, definitely has been dor the whole edition: that's why 1s don't auto-fail, too.

 

[tetrasodium]

I think this discussion is missing that bounded accuracy isn’t just applicable to the “skill system” in 5E—a term that I’m putting in quotes, as it’s pretty far from being a skill system in the sense of its use in D&D from AD&D OE where it was introduced through 3E. It is, at best, is a semi codification of the B/X system of “how to adjudicate things not in the rules.”

Bounded accuracy also applies to AC and to hit modifiers. It also includes the HP bloat of foes to account for the cap on AC. It also capped ability scores.

Looking at the effects of bounded accuracy solely within the frame of skills is really missing the forest for the trees.

Failing to look at the system holistically isn’t a really useful way to judge its performance.

BA is pretty much applied to the skill system in 5e for the GM. Compare the 5e one size fits all 5-30 dc ladder to the ones in 3.x when it was negative ten to 43 empowered by bonus types & GM's best friend or even 4e where dmg2 gave complexity 1-5 challenges & level based easy/medium/hard DCs in the spoiler

Spoiler: here

That's where so many of the problems in 5e's skill system come into play now that the GM lacks suitable tools to do anything but play the role of Pip in twilight zone's a nice place to visit episode.

 

[clearstream]

Then I blame being tired and distracted, but see below.

I could have made it clearer, for sure. Everyone else had been speaking about modifiers up to that point...

Should we really be considering the die as "+1 to +20"? That's...kinda giving the single biggest contributor here. As in, almost half of the absolute maximum someone could get, after ultra-hard hyper-focus on skill bonuses and nothing else. The +33 number is much more useful for most people, since everyone has always been able to crit in D&D, as far as I'm aware.

It makes sense to go with +33 if that is easier for others else to follow!

 

[Staffan]

Please explain exactly now you think that one hunan gm in real time can juggle the job of writing estoryboarding competence porn like leverage for skill checks after gm accessable quantum tools like DM's best friend were removed when they do not have full control like the writing team of your target example show.

If no single GM can meet the expectations of a skill system within the constraints of actual play, that skill system is an obvious failure

Present a variety of challenges. To go back to picking locks as an example: don't expect an unguarded door to be an obstacle if you have a lock-focused rogue around. Post guards that keep the rogue from accessing the door unseen, and will have to be distracted. Have patrols that will interrupt the rogue searching the now-no-longer-locked room. Put the locked room somewhere where it's hard to get to, so you need additional checks for Acrobatics and/or Athletics.

I think we may have a fundamental disconnect here. What do you think the purpose of a lock in an adventure is? To me, the main purpose is to make a lock-picker look great. If there is no dedicated lock-picker, then the party might need to look to other solutions if they aren't lucky enough to be able to open it anyway. But the purpose is certainly not to keep the party from accessing the thing behind the lock. Locks are meant to be opened. Secret doors are meant to be found. Chasms are meant to be crossed. Books are meant to be read (unless they're just bulky treasure).