D&D 5E Proficiency vs. Ability vs. Expertise

[Garthanos]

"Too much", "too much"

I think its safe to say that for most campaign,lengths those are once-in-campaign odds, right?

I suppose one could literally have people forced into arm wrestling contests LOL. It may not have to be at the same activity nor in a row however. In general I suspect it is highly variable by campaign.

1 in 6 or 1 in 5 tests (accounting for my off by 1) in opposed or compared checks I suppose would be different depending on how often you have those situations

Does it feel off shrug the fact that two performances 2d20 are really already involved and sometimes random chance concurs with more deterministic results too for me says maybe not too much.

But also on the whole it is a simple house rule to make less variation for careful or less intense situations.

 

[Esker]

Did you know that the typical rogue with expertise in RAW (assuming some reasonable boosts to DEX over his career) needs only an average of 5.6 (4.6 ties) on his check to beat foes with a CR equal to his level (or in the case of CRs 21+, when he is level 20)? Let's round that up to an even 6, so there is only a 25% chance the rogue will be noticed. This is looking at over 2000 foes, by the way.
...
Oh, and this assumes we grant every foe proficiency in perception... which is incredibly generous since many creatures and NPCs don't have proficiency in perception. So, really, his needed average roll is even lower...
...
Now, take away expertise, and that number increases to an average of 8.4 to beat passive perceptions against equal CR targets (with the same assumptions of every foe getting proficiency in perception AND no buffs, etc.). Rounding up to 9, there is a 40% chance to notice the rogue.

Considering the assumptions, that is more reasonable to me, especially when you consider his odds will be even better against most foes in most circumstances.

A couple things: first, if I'm a stealthy character, there is no way I'm risking going off and scouting ahead if I have a 40% chance of being spotted. Even 25% is dicey. So I'd need to have expertise as is to even risk trying. I suspect this is a case of a common probability fallacy, which is conflating "less than even odds" with "very unlikely".

Second, if there is more than one monster around, the only one that matters is the one with the highest passive perception. So that increases the chances. As does any monster making active checks. Granted if there's more than one, they probably individually have lower CRs, but still, combine multiple monsters and active checks, and it actually gets really risky to sneak around, even with expertise as written. But we already covered that.

Third, if I did grant that these stealth rates are too high, why is that a problem with expertise and not a problem with passive perception?

 

[Esker]

The GM has lotsa ways to resolve situations in 5e - if he gives it a 1 in 6, is it your place totell him "no"?

I'm not talking about the GM's prerogative to set the difficulty of a task; I'm coming from the GM's perspective here. For static checks vs DC, they can always tune the success rate by increasing or decreasing the DC. But for the opposed check mechanic, they don't really have any levers available to them, apart from granting advantage or disadvantage to one side. I love advantage and disadvantage as a mechanic in general, but using it on an opposed check merely because one side has a higher modifier would be really clunky.

The 2d10 mechanic makes opposed checks work more like I'd expect them to as a GM without messing with modifiers. Even better, leave both types of rolls available as options. Then the GM can call for 2d10 when they want skill to play a greater role, and 1d20 when they want luck to play a greater role. They could even have reasons to have one side roll a d20 while the other rolls 2d10.

 

[Maxperson]

There certainly should be visualized reasons behind the die rolls however it looks like someone who feels (subjective word) that the variation in performance caused by d20 exceeds what feels good/believable and or seems too much to emphasize fluke occurrences too much when one has a more steady controlled performance context. Allowing players to optionally make careful checks seems entirely low impact house rule to me like 3d6 or even 7 +d6 or 6+d8, 5+ d10, or similar if the circumstance seems reasonable (you could allow 3d6 to have a crit chance).


How often does purely random chance determine the results is the complaint and that isnt found by computing a mean/median, it is how often the die rolls vary more than determined factors however that too isnt good enough as arguably it only feels strange when random chance doesnt agree with a more deterministic result which makes it slightly different here is the formula for opposed d20s

n(n+1)/2 * 1/400

for a range where n = 11,+1 vs +8 = 16.5 percent
for a range of where n = 12, +1 vs +7 = 78 out of 400 (19.5 percent )

this is how often extraneous factors are assumed to interject.

Admitting my math might be bad this morning...

So admission or not, you're clearly a math guy.

What would the results be if you gave advantage to people who had +3 higher in opposed rolls, and again at +5 advantage? Perhaps advantage would work better than 2d10 or 3d6.

 

[Mycroft]

I think a 17th level rogue with a +17 grapple check vs. most monster's grapple check being sad; as most monsters are not even proficient in Athletics or Acrobatics, let alone double the number, is lame.

Seems like lazy design they hoped would just wash out.

 

[Esker]

So admission or not, you're clearly a math guy.

What would the results be if you gave advantage to people who had +3 higher in opposed rolls, and again at +5 advantage? Perhaps advantage would work better than 2d10 or 3d6.

Here are the numbers for all four methods (both 1d20, d20 with advantage vs 1d20 w/o, both 2d10, and both 3d6), rounded to whole percentages:

mod delta	d20	d20 adv	2d10	3d6
0	48%	64%	47%	45%
1	53%	69%	53%	55%
2	57%	74%	60%	64%
3	62%	78%	66%	72%
4	66%	81%	72%	79%
5	70%	85%	78%	86%
6	74%	87%	82%	90%
7	77%	90%	87%	94%
8	81%	92%	90%	96%
9	84%	94%	93%	98%
10	86%	95%	95%	99%
11	89%	96%	97%	99.5%
12	91%	97%	98%	99.8%
13	93%	98%	99%	99.9%
14	95%	99%	99%	99.99%
15	96%	99%	99.7%	99.998%

So, not surprisingly, giving the person with the higher modifier advantage at some threshold creates a major jump, and gives well above the 2d10 win chance up until the +8 or +9 range, after which it's almost identical to having both parties roll 2d10. The 3d6 method catches up to giving the better mod advantage at a difference of +4 or +5, then then rapidly hits the ceiling from there.

 

[Esker]

I think a 17th level rogue with a +17 grapple check vs. most monster's grapple check being sad; as most monsters are not even proficient in Athletics or Acrobatics, let alone double the number, is lame.

Seems like lazy design they hoped would just wash out.

Something like half of the monsters that present a challenge to a 17th level party are too big to grapple, fwiw. Also a raging barbarian with maxed strength (which is more common than a rogue with maxed strength) who is just proficient in athletics (for a +11 mod) has almost the same chance of success as the expert rogue (which you can see in the table I posted above). And the barbarian can much better afford to be the obvious target for whatever terrifying thing the party is fighting at 17th level.

 

[Swarmkeeper]

I think a 17th level rogue with a +17 grapple check vs. most monster's grapple check being sad; as most monsters are not even proficient in Athletics or Acrobatics, let alone double the number, is lame.

Seems like lazy design they hoped would just wash out.

Hmm... if a player with a Rogue has invested in Athletics proficiency and spent 1 of 4 Expertise on Athletics and used ASI to get Strength to 20, well, they’ve earned it. Your edge case “problem” really isn’t, IMO. Heroes gonna be heroic. Won’t ruin the fun at our table if that’s what the player really wants out of their Strength build Rogue. A DM has many many tools at their disposal to bring the challenge to said PC and their party.

 

[DND_Reborn]

Alright, fair enough. In the spirit of talking more about changes that might better achieve your stated goals, rather than the spirit of criticizing the need for the change in the first place, let me see if I can accurately characterize what you want to get out of these changes, and also summarize, as a rogue fan, and a fan of maintaining game balance more generally, what side effects I would want to avoid. And then I'll see if I can offer any suggestions that manage both.

The Goals:
1. Make skill proficiency a bigger factor compared to ability mods, when rolling skill checks.
2. Maintain uncertainty in outcomes, avoiding foregone conclusions resulting from modifiers that get too big

Side Effects to Avoid:
1. Ripple effects that cause aspects of the game unrelated to skill checks, to require major rebalancing
2. Substantially changing the relative frequency with which skill checks in general succeed or fail
3. Cutting into the rogue and bard's distinctiveness as skill monkeys

Before I suggest other solutions, is this a fair characterization of the priorities here? I get the sense that you may not be fully on board with my #3, so if you have an active goal of reducing rogues' and bards' ability to stand out from other classes in the skill domain, then we probably indeed have nothing further to discuss, since I'm not interested in ideas that make rogues just different kinds of fighters, or bards just different kinds of sorcerers. But if you are receptive to ideas that both achieve your goals and avoid my pitfalls, then I'm interested in continuing to think about it (I have an idea brewing already, but want to do some math with it before presenting it to see whether it does what I think).

Great! I have to work tomorrow (mandatory overtime, *yuck*), but let's make some magic happen! I'm sure together (and with the occasional idea from others) we can create something that works.

Goals:
1. Yes. My premise is that experience (and the "training") that goes along with it, reflected by proficiency, should be the largest factor. Initially, I was thinking roughly along a 50/25/25 split, so in theory expertise in a skill could match ability. My current idea is +8 max prof, +4 max ability (18 cap), and +4 max expertise. This is nearly the same potential as RAW at +17, and a slight boost over the +11 for non-expertise with a +12 potential.
2. Yes again. Although stealth has been my scapegoat in all this, as I said the same problem can exist with any of the skills. I have no issue with a rogue being as good as others, I just fail to see why they should have to potential to be better aside from a purely "let's give them something" idea.

Avoids:
Yeah, the all are good things to avoid. I don't want to detract from rogues/bards as skill monkeys, but would rather find some other way to represent the idea than just boosting their numbers.

A couple things: first, if I'm a stealthy character, there is no way I'm risking going off and scouting ahead if I have a 40% chance of being spotted. Even 25% is dicey. So I'd need to have expertise as is to even risk trying. I suspect this is a case of a common probability fallacy, which is conflating "less than even odds" with "very unlikely".

Second, if there is more than one monster around, the only one that matters is the one with the highest passive perception. So that increases the chances. As does any monster making active checks. Granted if there's more than one, they probably individually have lower CRs, but still, combine multiple monsters and active checks, and it actually gets really risky to sneak around, even with expertise as written. But we already covered that.

Third, if I did grant that these stealth rates are too high, why is that a problem with expertise and not a problem with passive perception?

Well, remember those numbers were without buffs, etc. against targets considered equal to you. Since more often than not your targets will be less than you in ability, etc. and when high risk is present you will probably have buffs of some nature. So, that 40% isn't "worst-case scenario", but it is close. True, multiple targets can make it more difficult, but unless it is active, once the rogue's roll beats the passive scores, he is in the clear.

Which brings #3, the problem with passive perception. Yeah, there is a problem IMO. I am not sure what to do with this one. I like the idea behind it, but not how it functions RAW. We make it really "passive" in that the DM uses it only when the player isn't specifying they are looking for something. I can explain it further if you want.

Anyway, ultimately I want to keep skill modifiers to +11 or 12, similar to attack rolls and saves. So, my current +8 prof and +4 ability works, but then removes expertise completely as a numbers boost. The idea of advantage gives better results, but keeps the max to the limit. Allowing expertise to function in other ways (make a check as a bonus action instead of an action, etc.). If the maxes are the same, then I am perfectly happy that a rogue expert in arcana can do as well as a wizard with max INT. Then I wouldn't have to grant expertise to other classes to get rid of the fluff idea of rogues simply potentially being "better" just because they can. This was one reason why i like expertise allowing a rogue to use the better of proficiency or ability, doubled. This works ok with RAW proficiency and ability, which would keep me from having to adjust things to my +8/+4 idea.

Well, that is it for now. I need to get to sleep since i have to leave for work at 7 AM. Sigh.. Looking forward to your thoughts.

 

[Mycroft]

Hmm... if a player with a Rogue has invested in Athletics proficiency and spent 1 of 4 Expertise on Athletics and used ASI to get Strength to 20, well, they’ve earned it.

I don't agree, not a big sacrifice, and certainly not deserving of inherently broken mistakes.

Tying grappling/shoving to skills was a big mistake, I feel.

5th Ed seems to have missed a few final edits or what-have-you (e.g. weapons table, monster proficiency based off of CR, etc).