How do we fix Bounded Accuracy? [Math], [D&D]

I've looked at it before, and I honestly thing that multiplying things by 2 (Ability mod is score-10, proficiency is just doubled) would solve all the problems for skills
>maximum level 1 bonus goes from +5 to ~+11, so there's actually a point in specializing at low levels
>level 20 specialist has +22, meaning they'll always succeed at what an untrained person could attempt
>even just the skill at level 20 is +12, which is enough that they always succeed at what an average person does half the time

You'd need to change Expertise (maybe just have it give advantage), and the thing that makes bounded accuracy good (combat never being a complete lockout by one side) falls out if you do this.

Simpler option that gets around the latter: using a d10 for skill checks. The decrease in granularity isn't too significant, since margin of success is basically not a thing, and it does the same thing to the odds of a given level and proficiency being able to succeed as the above.

I don't understand why it would be removed, I mean you have passive perception for fuck's sake that is literally just that. Obviously contested checks or other stuff with so many variables like Persuasion wouldn't have them, but if it is something your characters should be able to do, then there should be a routine check allowed.

>Bounded Accuracy
>a problem
>not a solution
But bounded accuracy is good

Unlike 3.x, everyone has a bunch of skills, and rogues only have 2 more than fighters.

>It's a houserule.
Oh. Well, that explains that.

Yeah. Passive checks are for the DM to roll against you in secret, not for you to autosucceed.

>Double the math
That would be a bit Overkill, but I can see it working for skills. I suspect it would fuck up combat though.

>Use a d10
That may mean training matters more, but it wouldn't fix the problem of trained characters being bumbling incompetents.

Maybe in combat.

Out of combat it's a swingy bumbling Clusterfuck.

Trap option is lazy design, but is not intrinsic of 3.5.

Obviously you'd adjust the DCs if you stepped down to a d10 (now that I think about it, a d12 would probably be better, since there's a chance of a +0 rolling something that overlaps with the +11's chances), such that the average person succeeds 50% of the time at a DC 5 (technically it would be 60%, but that's a minor note). In this model, the DC 11 is impossible for someone with no talent or training, but is a 50/50 shot for a level 1 specialist.

And yeah, doubling the math would certainly fuck up the one good thing 5e's combat has going for it, which is why changing the die used for skill checks would probably be better.

I don't know, before bound accuracy skills checks could get crazy.

I don't understand this meme.
Not thinking it's good, I mean posting
>but bounded accuracy is good
and nothing else on every discussion of the topic.

>meme
No user, bound accuracy is good. It truly is.

>Not intrinsic
I'm not talking about 3.x as the hypothetical game that could have been, I'm talking about the extant game that is.

3.x is filled with trap options that are a nightmare to slog through.

But being able to autosucceed on routine checks you're skilled at, and to know exactly what that amounts to (because you could print out the skill DCs and know if the task you're attempting is easier than what you could take 10 for) was one of few areas where PF was better than 5e.

I'll grant you, social skills had retarded DC examples - it was mind control, no save, full stop.

I'm not that familiar with 5e, but one way to increase the value of a few points is to use different dice. 3d6 gives you a slightly smaller range, and a much steeper curve. With a +6, you have a 74.1% chance to beat a DC of 15, while a guy with a mere +1 only has a 16.2% chance. On a d20, you're looking at 60% vs. 35%. Using 3d6, there is a gap of 57.9% between the success rates of the two people, while using a d20, there is only a 25% gap. That's quite a difference.

I think we essentially agre, I just find trap options not such a big deal, you generally recognize them on sight. Even less on a full-ogl game.

>and a much steeper curve
Which is to say that it has a curve at all, which makes it much steeper than a flat distribution.

>DC10 is easy
>"Anything a normal person fails 50% of the time, is not easy."
The DCs aren't for normal people, they're for PCs. Are you really getting hung up on them naming DC10 as "easy". That's crazy.

They're (painfully obviously) based on what a level 1 character who's focused on that skill can do.

I'm a new player to 5e, can you explain what bounded accuracy is?

Replacing d20 roll with dice pools would be a good start

Basically they cap DCs, ACs and how high your Ability scores can be ect.

That way there's less level disparity, and shit doesn't get too ridiculous as far as trying to hit an enemy or succeeding a skill check.

No, that would make hitting average numbers more likely, but making hitting high numbers less likely. Using a dice pool would be more of a double edged sword than anything.

>but making hitting high numbers less likely
Adjust your high result floor accordingly. Dice pools certainly reduce probability that skilled characters fail simple rolls better than highly volatile single roll. Dice pools are simply better than d20, face it.

>Adjust your high result floor accordingly.
You might as well just let people auto succeed without a roll.

>Dice pools certainly reduce probability that skilled characters fail simple rolls better than highly volatile single roll.
You sound like someone salty because they made some bad rolls last session. It a dice game, there's supposed to be luck involved.

>Dice pools are simply better than d20, face it.
Absolutely not.

can they put a cap on your hit modifier?
I'm currently sitting on +9 at lvl4 and I'm using the precision skill from fighter to get +1d8 to hit.

>>Adjust your high result floor accordingly.
>You might as well just let people auto succeed without a roll.
We are talking about top perfomance here. Dice pools scale it better and provide a way to measure degree of success.>You sound like someone salty because they made some bad rolls last session. It a dice game, there's supposed to be luck involved.
I'm not and my last session wasn't d20 game of any kind. It was d100.
>Absolutely not.
Denial isn't just a river in Egypt, user.

>can they put a cap on your hit modifier?
In a round about way they do. Your hit modifier is your Dex or Str mod + Proficiency + whatever else. Proficiency modifier caps at +6. Your ability score modifier have a soft cap of +5 and a hard cap of +7. So it should cap at 11 or 13 + whatever other bonuses you have.

td;dr The numbers that make up your attack modifier have caps

A single dice has flat chance for every result. A dice pool creates a bell curve making "middle" numbers more probable and "high" and "low" numbers less probable. Good if you need a "middle" roll, bad if you need a "high" roll.
>Hurr durr, lower the DC so you don't need the "high" rolls
You could do that with a single dice. But doing that with dice pools so every roll needed is a "middle" roll kind of defeats the purpose of rolling. Yeah, there's a slight chance you lose, but you'll usually succeed... on what would normally be difficult. Why do that? You would practically auto succeeding you're players.

You know people who don't have fragile egos get bored when games are too easy.

Dice Pool combined with Bounded Accuracy actually makes for a very interesting probability spread.

The reason for Bounded Accuracy is to make granular modifiers less damning, and also less important than the die roll.

Dice Pools make the probability of the rolls matter less due to their weightedness.

Add those together and you're looking at a system that is nearly predictable, not that that's a bad thing. I'd love to try out 5e with 3d6+2.

The only thing that would be finnicky then however would be the advantage/disadvantage mechanic. Instead of giving roughly +/- 3 it would be giving something probably closer to +/-1.4 or so. Maybe in that case, advantage/disadvantage could just become a flat +/- 3 modifier to the roll. That's what I'd do at least.

Ah. I recognize them fairly quickly, but it still means sifting through a mountain of sand by hand looking for stones. It's slow.

And where everything has obnoxious prerequisites, you need to plan several levels ahead, which makes things even slower.

If you're failing 15% of the time, and you're good at something, that's definitely not easy.

Attributes: +1..+5.
Proficiency: 1+RoundUp(lv/4)
Expertise: Double proficiency bonus
Magic Items: +0..+3 (and no guarantee you will have one).

Much slower dice bonus advancement.
But it also still starts with really low numbers, which results in the issues OP mentioned, regarding skills.

>If you're failing 15% of the time, and you're good at something, that's definitely not easy.
No. If I shoot 20 baskets and I make 17 of them, I think I'm doing pretty good.

If you're shooting 20 baskets and you get 17 of them, you are doing pretty good.

But if it was "easy" you wouldn't be missing more than 1/10.

>But if it was "easy" you wouldn't be missing more than 1/10.
That's making 18 out of 20 baskets.

You are officially retarded.

Yes. Yes it is 18 out of 20 baskets. What's your point? You've got to set a minimum somewhere.

An easy task is one you basically cant fail. I can chalk

different guy here.
heres my analogy. free throws are easy. 85% puts you at 30th in the NBA. fuck your stupid logic, im going to another thread.

I've heard this take 10 or take 20 rule before but I've never seen it I the rule books. Is it a home brew fix or is it jus really deep in the dmg?

If you have the opportunity to take 10 or twenty, why should you have to make a check at all? In that case, just say you need +X modifier to pass.

It's a rule from 3.5 that many people haven't realized doesn't exist in 5e.

Take 10 is not very useful when you don't have a big list of example tasks so that you know the approximate difficulty of anything you might attempt, withing like 3 pts. In 3.x you have such a list. In 5e you don't. Basically, when you know a 10 is good enough, and you're not under pressure, you don't roll, you just take 10.

Take 20 is for when you can attempt something over and over again and all it costs you is time. Rather than rolling until you get a 20 to see if you succeed, you take 20x as long, say you got a 20, and see if that's enough to succeed or not.

Take 10 is the autosuccess.

If you're not being attacked or chased, you can take 10. It's the "I line up the shot and do it right", vs shooting from the hip because you don't have time to line up the shot.

You succeed if your skill is good enough, you fail if not. Unless you know the DC, you have to guess whether to roll or take 10. Thus, using take 10 for all but the simplest tasks only really happens if the player knows/has a cheat sheet of relevant DCs and therefore has a really good idea of if taking 10 is a good idea.

Don't forget the 'add proficency when using the help action' feat, it's an easy boost for every party.
The maximum is +30, dc40 isn't hard (with advantage), dc50 is the max,

And I'm saying that based on those odds, free throws are not easy.

Driving a standard without stalling the car or crashing, is easy.

Balancing on an 18 inch wide platform without falling, is easy (an official DC 10 example).

Making a free throw is not easy. It's easier than many other tasks in basketball, but it's not easy overall.

Which feat is that again?

Why not have a prestige-based game, where a character changes classes every five levels into different prestige classes?
For your first class and every time you 'promote' into a greater class, you get a +3 bonus to that class's skills. You could have magic powers based on spell points, which classes gain by level, instead of spell levels, and if there was a cut and dry way to create a spell (as well as spell examples already given by the book) DMs could create spells appropriate to the campaign. Spell points would be gained on a level-by-level basis, and martial dice or the equivalent could be used in much the same way. The higher your number of spell points, the greater the spell you can cast. Hell, add a third increasing feature like Expert Dice for things like rogues and rangers that help out with certain skill checks.
If you wanted to go down the cashgrab route, new books could be used to give new features that you could attach to spells, and weapon attacks for the cost of new points. Like some sort of horrid bastardization of 4E and GURPs.
I really need to stop going on Veeky Forums when tipsy. Every time I do I turn into an idea guy for a fantasy heartbreaker.

The 4e/Star Wars Saga Edition curve was literally perfect for the d20.

Everything advances at 1/2 level (+ stat when applicable), trained skills get a +5, expert skills get another +5 (this last one is a SE thing).

Nigger, you are forgetting that dice are supposed to represent variance and chance. Sometimes you hit 19 or 20, sometimes just 16 or 17. People aren't machines and usually don't act in perfect laboratory conditions, some variance is to be expected and desired by the resuliton system. Otherwise you could forego rolls completely and say "If you have a 7 or higher, you pass"
Incidentally that's also why there should be a small chance that a trained person fails an easy task, because that happens, sometimes. No, get that natural 1 horseshit out of here, that's terrible.

I've applied a generic -4 across the board on things a character is untrained in.

So the gap between trained and untrained starts at 6, or 30%, which is quite a bit more substantial. It also effectively shifts their range of rolls from 1 to 20 to -3 to 16.

Realistically people tend to become proficient at things their best stats line up with. For a character out of the newbie levels, I consider +7 (4 from stats, 3 from prof) about normal for the stuff they'll be trying for the group, which means they'll reliably hit dc10 with 90% success. Someone not proficient will tend to be more about -2 overall (2 from stats, -4 from nonprof), meaning they'll hit dc10 about 45% of the time. So the trained user is "almost always" and the untrained user is "maybe", which is where I want it. Shifts that around a few percent depending on the power level of your game. Also this puts the disparity between trained and untrained at 9 points in actual use.

DC5 has a purpose, as the untrained user can still fail them a third of the time, while the trained user doesn't even need to roll. Going up to DC15 puts the trained user in 2/3 range and the untrained in 1/5. Going up to DC20 puts the trained user in more uncertain 2/5 while the untrained user simply can't. DC25 is for heroic efforts that even the trained user drops to about 1/6.

What can you expect of an autistic retard?

Bounded accuracy is a legitimately good mechanic, 3.5/PF might be more up your alley, OP.

Already addressed that comment.

I like Pathfinder way better for skills, and like 5e better for everything else.

No, it's not. It's directly responsible for HP staying at 3.5 levels but damage being half of that if we're being generous. It's also responsible for the horseshit that is a top of the line level 20 non-Rogue or Bard failing a check then some untrained, untalented chucklefuck succeeding on it.

4E's skill system was far better than this.

Considering that a level 1 PC with proficiency in a skill should have at least +3 to +5, the DC 10 task should be pretty easy. The DC assumes that you're doing it in a rush, anyways.

In the way of "useless-expert", have you heard of Expertise?

Yeah, that's the reason I'm inclined to want to fix the skill math.

Also, you're right about HP getting pretty bloated. But what really matters in terms of hp is the number of attacks it takes to kill something, and the number of hits you can survive.

Out of curiosity, how would you fix 5es math scaling?

Move the attribute caps and make ASIs independent of feat progression, give everyone Expertise in a few skills, make crits actually double damage, remove BM Fighter, then make their dice and maneuvers a standard part of *every* martial to a lesser or greater extent(ie: War Cleric/Valor Bard get less dice compared to a Paladin or Ranger or Fighter).

That's just if I wanted to stick to stuff that's easily done with existing 5E content. If I was making it for me, I'd jettison the whole fucking game and make ToB 2.0.

I'm interested in hearing more...

Where would you move attribute caps to?
>ASIs separate from feats
I was leaning towards this as well. Make feats into 2 Pt. Things, and the ones that normally give a +1 stat, you don't get the +1, and it's worth 1 Pt. Give points spread evenly, maybe 10/20lvs or so.

>Give everyone expertise
Then why not just drop it and increase the progression speed?

>Maneuvers and dice to all martials
I definitely agree regarding maneuvers. Why for dice?

Let's presume, for arguments sake, we're making 5e work, not starting over from scratch.

>Where would you move attribute caps to?
26. 5E may say that magic items aren't necessary, but that is a blatant lie if you take one look at high CR enemies and their AC and attack bonuses. +3 to hit and damage, and AC for Dex users, solves part of that issue.
>Then why not just drop it and increase the progression speed?
Because you only need to ensure that more classes get Expertise to turn it into the two-tier training system that it functionally is for Rogues and Bards.
>Why for dice?
Because that was the original concept for them and it's a far better idea than what we ended up getting.

>26
So, bring back WBL, or scale up attributes more?

>just have Two tier training
But that doesn't really address the bad chumpiness of experts until the highest levels, and it also drastically increases the effective per-level scaling, up to ~2+half level. That's why I suggested a flat bonus plus a scaling level bonus - like 4e.

>Playtest spread the dice around more and it's better than what we have now.
Fair enough.

>Where would you move attribute caps to?
They should cap at 36. Just like levels.

Go home old guy. You're drunk. Take your racial level caps, staggered XP charts, race as class, and descending AC with you.

Hey, 4e capped at level 36! I'm still cool!

I'm no 4rrie, but I'm pretty sure 4e capped at 30.

For players, but Bahamut was the highest level monster, at level 36.

Ah. I hadn't considered enemy levels

Your mistake was thinking 5e still clung to baggage from 3.5/PF severe autism edition.

>He assumes everyone is "le samefag"
>He thinks having odds that actually make sense is somehow PF baggage.

That is the worst possible name I think you could give something. So you're proficient in a skill, and you also get Skill Proficiency on top? Why would you name your secondary bonus the same thing as the first bonus?

If you have to invest so much in skills to be able to use them, you're going to have trap options. It's not even going to be up to the game to decide what they are.

Your STAT for a check can only be increased or decreased once. That is: only one bonus applies at a time. Bring back granular bonuses.

When I wrote it I was thinking of having proficiency be a flat +8, and then have level scaling apply to *everything*. But making adjustments to not drastically alter combat and such was getting too finicky, so I reworked the bonus to only apply to skills. I agree it's a bad name if proficiency bonus is still a thing on top of that.

What, like stackable advantage?

That still doesn't help how little difference training makes.

I think the central issue is that D&D isn't realistic enough.
There's no point in the dice/stat matrix where a trained character nearly always succeeds where an untrained person nearly always fails.
For an untrained person who just learned what a plane is the moment before isn't going to able to randomly pilot a plane successfully. A trained pilot is almost never going to crash a plane, disregarding external factors.

You realize that on skill checks where a person should auto-succeed, you're supposed to not roll at all, right?

What you mean PFs skill point system?

Yeah. When I play PF I assign skillpoints in order to hit a certain amount of consistency. Some skills I level until +5, others +7, some +15. Contested skills I raise every level.

I agree, people do fuck that up. It's spelled out for you pretty plainly though. The DCs are laid out in the book, and it's easy to know when you can take 10 or take 20, and what tasks you can do that with. But it means bringing a skill DC cheatsheet for the skills you use, and keeping it as part of your character sheet.

I'm not advocating that for 5e. However, I do prefer that to the mess that 5e has, which is that nobody is competent (unless you override the published DCs and set new ones), and the difference between a trained character and an untrained one is next to nothing unless you are a high level, or a middling level with expertise.

Proficiency = +5, nonscaling, add 1/3.5lv to all checks, and expertise doubles level bonus (or gives a flat +3) seems the way I want to go, to start.

The gap between trained and untrained is about the same as the upper levels, the scaling is nearly the same throughout. Then I can just adjust from there.

Which really, 5e's skill checks blatantly go by the opposite because it's a game and they want every player to be able to fail or succeed a check more than they want to be realistic. It's not really a place where the game failed as a place where they went in a direction you didn't want.

So, how do you do that for tasks where a trained person should autosucceed, and anyone who isn't trained would have a good chance of failure?

At what point do I say: Brian, you don't have to roll, but everybody else make me an acrobatics check not to fall down.

What action is this where the skilled person has absolutely no chance to fail, but everybody else has a high chance of failing? What's so easy a braindead half-asleep moron could do it if they just took a class on it the day before, but hard enough to warrant a roll in the first place?

For 5e:
When Brian has expertise in the skill and literally cannot fail, remembering that nat 1s are not autofails on skills?

When Brian is a rogue in general, and cannot get less than a roll of 10?

Okay. Yes. They deliberately set the DCs such that everybody has a chance to succeed at everything, and experts have an unusually high chance to fail shit that should be trivial for them.

The druid with points in Nature should 100% be able to do stuff with it that the rest of the party would likely fail.

And yes, I do not find the swingy wobblyverse math to be at all desirable. I do want more realism and competence with skill performance than 5e offers, as that other user mentioned.

Hence this entire fucking thread.

>It's not really a place where the game failed as a place where they went in a direction you didn't want.
This is a problem because in my 5e games I've wanted to do something I was trained in and still. Constantly. Fail.
It's not the untrained succeeded alone that is the problem, it's that being proficient in something still has a decent chance of failure.
I am still salty over being an orc barbarian and rolling

I'm seeing a bunch of you wanting to bring PF 's or 3.5's mechanics into the mix. If you can't see these systems are very poorly designed then there's no helping you and you're trying to shove horseshit mechanics into an actually well designed system.
But hey, all the more power to you.

Hmm. For some real world examples?
>Driving a standard without stalling for 10 minutes, for someone who has never driven and is receiving no directions.
>Writing a simple command line dice rolling app within an 40 minutes without help.
Et-fucking-cetera.

Exactly. Hence my problem with 5es math. Flattened progression, no significant bump up for training at the beginning, means you need to have expertise and enough of a bonus for that to matter to be good at _anything_

Those aren't really things you should roll for, they're just following instructions. Do you roll for grocery shopping?

Why the actual fuck do you want to roll for them? What the fuck, has Pathfinder completely rotted your brain or something.

Op here. Nobody has mentioned that at all.

People have however stated that "even Pathfinder's skill system is better than 5es skill system". I agree with them. PFs skill system has issues. It's overly fiddly. Some of the DCs are stupid. But it's at least *possible* in PF to be competent in a skill without being level 9+, with levels in rogue.

It's not about a love for skill points. It's about bonuses being abysmally low for trained characters in 5e.

Those are things where you'd automatically fail without training, so they don't really fit the criteria that well. You wouldn't even roll, you'd either have them interact with the car and have them see what happens based on what they try, or in the command line dice rolling app just have auto-fail without training because how could you ever hit upon that set of directions without training ever. Those are actions that are complicated, but not hard: dice rolling never comes into the question.

And you seem to assume that anything less than a +10 is a worthless skill investment.

3.5's problem was that the skills were incredibly poorly balanced against each other, that certain skills had breakpoints beyond which more points were not useful, that adding INT to skills known favored spellcasters and screwed over classes that couldn't use INT for anything, and that cross-class skills were shit. It's not the large numbers at all, it's the fact that some people were completely locked out of doing anything useful with their skill points while others got to jack off to having the god skills of Diplomacy and UMD.

Presumably because they're things your medieval knight has a good chance at failing, and there's some sort of story consequence for doing so. It was simply an example of a task you are unlikely to succeed without directions, and which you basically can't fail if you know what you're doing, because seemed to think there were no such situations in life where skill difference would be so relevant, so I gave him some obvious and simple examples.

You realize that 5e is a genre shift due to grittier, lower powered fantasy don't you?
If you want to break the system assumption, that's fine, but it means you're also completely misunderstanding the point of the system.

No shit your medieval knight would fail because the things wouldn't have been invented in his lifetime. But the fact in that context you need rolls for every little thing suggests you're autistic and need an autistic system, 5e isn't that and you're better off playing PF.

>So, how could this be fixed
Nigga, the stuff you said could literally just be fixed by the GM adjusting DCs for skill checks.

Yes, it is shifted to a lower power level.

And trained characters are less competent in skills than starting AFMBE characters,

A shift from super heroes to action heroes is fine. The math for skills is more like a shift from super heroes to 4th graders.

Giving different DCs for the same task, to different players? I'm sure that will go over well.