Bell curve versus flat die roll preferences

[MrStabby]

I have a mixed view. What I hate about Bell curve mechanics is that in practice it tends to shut players out from certain parts of the game. The non charismatic character will fail at diplomacy as getting an 18 on 3d6 is pretty much not going to happen. On a d20... yeah, its a tough roll, but go for it.

Likewise it can blow up small differences in characters or trivialise some investments. If you are hitting on a 4 a +2 on 3d6 won't make much practical difference, but if its a d20 then yeah, there is a not entirely trivial probability of failure removed. On the other hand, if you are needing a 16 to hit, then the next player who only has one smaller bonus compared to you is massively disadvantaged. As a DM I'd don't like such a system trying to force me to use a modest band of creatures where these characters can play a similar game. I like D&D 5e bounded accuracy and would like to keep it.

On the other hand bell curve systems have a nice, internally consistent way of handholding expertise. Something one character should certainly know, but is unlikely for another? What should the DC be? What if there is only 11 points of difference in bonus between them? No DC will meet that in a flat system. Now add the added complication of every other party member taking intermediate values. You can put an arbitrary cut-off on where either someone automatically knows or doesn't know or even both, but if you are having these cutoff points they can equally well seem arbitrary and antagonistic.

Basically, I think it comes down to whether you want to include PCs in that part of the game or for it to be a specialist area that needs investment in and you want to exclude non specialists.

I have seen in a few games that there are a lot of bonuses that are dice, not static. To keep to my familiar d&d 5e, there are things like bless, bardic inspiration, dark ones own luck, various Eberron races on skills and probably a whole bundle more not on my list. These do result in a bit of a middle ground between the two, with the d20 dominating but still making the probabilities of the absolute highest and lowest rolls much less than a flat system.

[stoutstien]

I have a mixed view. What I hate about Bell curve mechanics is that in practice it tends to shut players out from certain parts of the game. The non charismatic character will fail at diplomacy as getting an 18 on 3d6 is pretty much not going to happen. On a d20... yeah, its a tough roll, but go for it.

Likewise it can blow up small differences in characters or trivialise some investments. If you are hitting on a 4 a +2 on 3d6 won't make much practical difference, but if its a d20 then yeah, there is a not entirely trivial probability of failure removed. On the other hand, if you are needing a 16 to hit, then the next player who only has one smaller bonus compared to you is massively disadvantaged. As a DM I'd don't like such a system trying to force me to use a modest band of creatures where these characters can play a similar game. I like D&D 5e bounded accuracy and would like to keep it.

On the other hand bell curve systems have a nice, internally consistent way of handholding expertise. Something one character should certainly know, but is unlikely for another? What should the DC be? What if there is only 11 points of difference in bonus between them? No DC will meet that in a flat system. Now add the added complication of every other party member taking intermediate values. You can put an arbitrary cut-off on where either someone automatically knows or doesn't know or even both, but if you are having these cutoff points they can equally well seem arbitrary and antagonistic.

Basically, I think it comes down to whether you want to include PCs in that part of the game or for it to be a specialist area that needs investment in and you want to exclude non specialists.

I have seen in a few games that there are a lot of bonuses that are dice, not static. To keep to my familiar d&d 5e, there are things like bless, bardic inspiration, dark ones own luck, various Eberron races on skills and probably a whole bundle more not on my list. These do result in a bit of a middle ground between the two, with the d20 dominating but still making the probabilities of the absolute highest and lowest rolls much less than a flat system.

Not to get to far into the weeds of ludology and design choices but I think a problem that most modern game design has in common is the being over reliant on the critical path to the point that it tries to predict the players choices (in game and within the meta portion of choices for their PC) and then maps out the "math" backwards.

The critical path is a good thing to have because it can give you a structure but over loading it will mean adding in more and more gates. Gates inevitably mean some form of check or test.
falling back on the resolution mechanics as a mean of progression within parameters in which the players lack agency (usually the critical path) is one of the leading causes of the Modern player mentalities of optimization and/or min/max, and an over reliance on meta knowledge of mechanics.

[GloatingSwine]

I dunno, I think min/max optimisation mostly comes out of vertical progression.

If the game has a long ladder of vertical progression for characters that impiles a rising water level of progression for the things those characters will encounter, and any automatic bonuses they'll get as they climb the ladder will just be the basics they need to keep their head above that water.

So the way they feel like they're getting ahead of the system is to minmax.

In games where the gap between being useful at something and the best at something is narrow the design naturally can't have that rising water level, and so the players can more freely choose to broaden their characters instead because there's more value in being able to do more types of things just well enough.

D&D, because it's such a vertical progression game, encourages you to be a specialist so that you can beat the rising water level.

[stoutstien]

I dunno, I think min/max optimisation mostly comes out of vertical progression.

If the game has a long ladder of vertical progression for characters that impiles a rising water level of progression for the things those characters will encounter, and any automatic bonuses they'll get as they climb the ladder will just be the basics they need to keep their head above that water.

So the way they feel like they're getting ahead of the system is to minmax.

In games where the gap between being useful at something and the best at something is narrow the design naturally can't have that rising water level, and so the players can more freely choose to broaden their characters instead because there's more value in being able to do more types of things just well enough.

D&D, because it's such a vertical progression game, encourages you to be a specialist so that you can beat the rising water level.

I used to think this until I asked myself why the explosive progression (for some aspects is larger than exponential) became ingrained.

It's a workaround for the style of adventure or module writing and an over focus on balance and faux progress.

If they slapped a troll in a cave in the game they felt like there needs to be some form of gate to prevent the players from encountering it before it's 'fair'. Once the idea of fair or balance became more important than the why or how it inevitably lead to what we are accustomed to nowadays.

*Ironically this is 5e strong suit because it doesn't actually care about balance or fairness past window dressing. It's a decent chassis because it doesn't rely on progression but it allows players to think it does.*

[GloatingSwine]

Trying to have balanced encounter design certainly reinforces the rising water level, but just having vague ideas about "here's a bunch of stuff you fight and do at low levels, here's stuff you fight and do at high levels, and here's some stuff in between" is, I think, enough to make people want to be as good as they can for their current zone.

[Quertus]

Is that something that happens in Battletech? Or is it something that happens in other system(s) you're comparing to Battletech? If so, which system(s) do you mean?

Not a Battletech issue. Not comparing to Battletech. Chocolate is yummy, because sweet and chocolate. Dirt is yucky, because gritty and dirt. Giving examples of good and bad, not comparing the subjects of the comparisons.

Perhaps would have been clearer with the inclusion of phrases like “for example”.

[Pauly]

I have a mixed view. What I hate about Bell curve mechanics is that in practice it tends to shut players out from certain parts of the game. The non charismatic character will fail at diplomacy as getting an 18 on 3d6 is pretty much not going to happen. On a d20... yeah, its a tough roll, but go for it.

Likewise it can blow up small differences in characters or trivialise some investments. If you are hitting on a 4 a +2 on 3d6 won't make much practical difference, but if its a d20 then yeah, there is a not entirely trivial probability of failure removed. On the other hand, if you are needing a 16 to hit, then the next player who only has one smaller bonus compared to you is massively disadvantaged. As a DM I'd don't like such a system trying to force me to use a modest band of creatures where these characters can play a similar game. I like D&D 5e bounded accuracy and would like to keep it.

On the other hand bell curve systems have a nice, internally consistent way of handholding expertise. Something one character should certainly know, but is unlikely for another? What should the DC be? What if there is only 11 points of difference in bonus between them? No DC will meet that in a flat system. Now add the added complication of every other party member taking intermediate values. You can put an arbitrary cut-off on where either someone automatically knows or doesn't know or even both, but if you are having these cutoff points they can equally well seem arbitrary and antagonistic.

Basically, I think it comes down to whether you want to include PCs in that part of the game or for it to be a specialist area that needs investment in and you want to exclude non specialists.

I have seen in a few games that there are a lot of bonuses that are dice, not static. To keep to my familiar d&d 5e, there are things like bless, bardic inspiration, dark ones own luck, various Eberron races on skills and probably a whole bundle more not on my list. These do result in a bit of a middle ground between the two, with the d20 dominating but still making the probabilities of the absolute highest and lowest rolls much less than a flat system.

This has a lot to do with bonuses.
If you’re playing D&D by the time you’re mid level your bonuses become more important than your base abilities and die roll. It’s pretty easy to build a rogue with 8 CHA who is a better all round party face than a 16 CHA paladin by level 10. Likewise by level 5 or 6 a fighter’s STR stat is trivial in combat, so I often built fighters with STR of 12 or 13 if I was playing in a campaign with rapid leveling..

On the other hand if you’re playing Traveller, for example, then bonuses are small and don’t stack. In which case the die roll and base ability are still very important even with highly advanced characters.

I don’t think it matters if you’re using a flat d20 or a bell curve 3d6. As soon as the bonuses start overwhelming the die roll then the game changes and all the nuanced probability models fall by the wayside.
In D&D when a level 1 fighter hits a target they roll for weapon damage damage and add their STR bonus. When a level 20 fighter hits a target they do damage then roll for bonus weapon damage.

[NichG]

Hm, what about a system like the following. Character skills and attributes have a very granular 'scale' such that they never receive temporary bonuses or maluses directly to that scale number - any effects which impact those numbers at most set them to a fixed value, rather than adding/subtracting. Effects which do additively stack instead go to a secondary associated score which we could call 'fractions' or 'pips' or something.

Using a flat d20 as the base here, differences in scale count as +/- 3 per point of difference, but you don't add your scale into the roll - it always gets factored into the DC. Pips on the other hand determine the minimum (if positive) or maximum (if negative) number that can be obtained on the d20 roll. So if you have 5 pips from buffs, gear, situational stuff, etc, then a result of 1-5 counts as a 6. If you have -5 pips, results of 15-20 would count as a 14. You wouldn't be able to have more than +/- 19 pips from any sources - that would turn every roll into a 20 or a 1. There wouldn't be critical success/failure in such a system. When used to determine the extent of things (like damage, movement distance, etc), treat every 4 points of scale like a doubling, in the pattern 100%, 125%, 150%, 175%, 200%, 250%, 300%, 350%, 400%, ... Basically 2^int(scale difference / 4) * (100% + 25% * (scale difference - 4 * int(scale difference / 4)).

The way I imagine this, you'd have some broad categories like Body, Mind, Spirit that are pretty much set for a given creature category and aren't really variable through point buy kinds of things - the usual playable humanoids all have, say, 8/8/8 in this. Individual skills, characteristics, attributes, etc would just act like specializations, and each would have a fallback rule of either 'fallback to parent attribute' or 'fallback to zero' - equivalent of 'trained only'. If you wanted to separate Strength and Dexterity or something, they'd be subdivisions of Body with 'fallback to parent attribute', and under those subdivisions you could have individual uses of Strength or Dexterity or whatever. Anything that has 'fallback to parent attribute' gets increased/decreased from that parent attribute level, whereas anything that is 'fallback to zero' is increased/decreased from zero regardless of the value of the parent attribute. However, in either case, the scale of subordinate characteristics are limited to be at most, say, parent attribute + 3. So if you have Rock Climbing under Strength under Body, and a Body of 4, you might be able to support a Strength scale of 7, and a Rock Climbing scale of 10, and a specialization of say Chimneying under Rock Climbing with a scale as high as 13. Any points in a subordinate characteristic beyond that give you pips instead.

I think that basically achieves the goal of having consistent variance even if you have massive progression, while also allowing progression to be limited by not having ways to change the parent attributes beyond a certain point if that's what you want.

[Quertus]

stuff

Reading comprehension is my bane, but it sounds like this potentially solves my “I don’t need to roll to put on my pants… unless I’m drunk, or a small child” problem, no?

[NichG]

Reading comprehension is my bane, but it sounds like this potentially solves my “I don’t need to roll to put on my pants… unless I’m drunk, or a small child” problem, no?

Yeah, you could solve this a couple of ways. One way would be to say that some tasks aren't things which are in question - there are no real sources of uncertainty - in which case you just have to meet the scale of the task, and then you can do it. Or, even if you in principle 'roll for everything', if say the DC of a task at the same scale as your ability is 10, anything 4 scales below you is an auto-success, and 3 scales below with even a few pips of support (from tools, help, being able to take your time, whatever) is also an autosuccess. Similarly, no matter how much support you have, you wouldn't be able to succeed at something 4 scales above your own.

I'd personally tend to go with the idea of 'only roll in the presence of unknowns or external agency', because it makes sense to me that different contexts would have different variances - hitting a stationary practice dummy is both easier and also less variable than hitting a moving target, not just easier. So, you're sewing a sole onto a shoe for the thousandth time, in your workshop, and there are no disruptions or problems with the leather or whatever? Check scale, call it done. You're doing exploratory surgery where you don't exactly know what's going on inside the patient? Roll. You're doing routine surgery with the help of a realtime futuristic x-ray/sonic/neutrino/etc imaging booth so you know exactly what's there where you're cutting? No unknowns or external agency, so don't roll. But that's a preference and what makes sense to me.

[PhoenixPhyre]

Yeah, you could solve this a couple of ways. One way would be to say that some tasks aren't things which are in question - there are no real sources of uncertainty - in which case you just have to meet the scale of the task, and then you can do it. Or, even if you in principle 'roll for everything', if say the DC of a task at the same scale as your ability is 10, anything 4 scales below you is an auto-success, and 3 scales below with even a few pips of support (from tools, help, being able to take your time, whatever) is also an autosuccess. Similarly, no matter how much support you have, you wouldn't be able to succeed at something 4 scales above your own.

I'd personally tend to go with the idea of 'only roll in the presence of unknowns or external agency', because it makes sense to me that different contexts would have different variances - hitting a stationary practice dummy is both easier and also less variable than hitting a moving target, not just easier. So, you're sewing a sole onto a shoe for the thousandth time, in your workshop, and there are no disruptions or problems with the leather or whatever? Check scale, call it done. You're doing exploratory surgery where you don't exactly know what's going on inside the patient? Roll. You're doing routine surgery with the help of a realtime futuristic x-ray/sonic/neutrino/etc imaging booth so you know exactly what's there where you're cutting? No unknowns or external agency, so don't roll. But that's a preference and what makes sense to me.

Another aspect is interesting consequences. Even if it's uncertain, if there aren't interesting consequences for both outcomes (success and failure)? I prefer not to roll. Because you're taking table time with something you've decided is uninteresting. Pick the interesting branch (if either) and go with it.

[NichG]

Another aspect is interesting consequences. Even if it's uncertain, if there aren't interesting consequences for both outcomes (success and failure)? I prefer not to roll. Because you're taking table time with something you've decided is uninteresting. Pick the interesting branch (if either) and go with it.

Yeah. Ideally the way the system is structured would make this somehow obvious as to how to determine if consequences are interesting. This is sort of why I like 'roll to see how much something costs' rather than 'roll to see if you succeed' as a design element, though it didn't quite fit in this scheme...

Maybe something like, the most common way to get pips is to spend a relatively slowly-recovering resource, which can be done after the roll (even, say, hours after the roll), and a failed roll really means in all cases 'this is the price of that resource for you to do this thing'. So in no case can you ever 'try again' on any roll in the d20 sense - it costs what it costs.

I know people who don't like 'roll to determine things about the world' would hate that though.

[PhoenixPhyre]

Yeah. Ideally the way the system is structured would make this somehow obvious as to how to determine if consequences are interesting. This is sort of why I like 'roll to see how much something costs' rather than 'roll to see if you succeed' as a design element, though it didn't quite fit in this scheme...

Maybe something like, the most common way to get pips is to spend a relatively slowly-recovering resource, which can be done after the roll (even, say, hours after the roll), and a failed roll really means in all cases 'this is the price of that resource for you to do this thing'. So in no case can you ever 'try again' on any roll in the d20 sense - it costs what it costs.

I know people who don't like 'roll to determine things about the world' would hate that though.

Determining if something is interesting is a quintessentially fact-and-table-bound problem. What's interesting at one table is boring at another. So outside of general guidelines (of the "don't bother rolling if you can effortlessly try again and time doesn't matter"), I prefer if the systems don't bother trying to enforce or "encourage" this. Because most of the time it just ends up being something we (as a table) have to fight against. I dislike "opinionated" systems--I prefer systems that provide tools but don't try to force one way of using them over another.

[NichG]

Determining if something is interesting is a quintessentially fact-and-table-bound problem. What's interesting at one table is boring at another. So outside of general guidelines (of the "don't bother rolling if you can effortlessly try again and time doesn't matter"), I prefer if the systems don't bother trying to enforce or "encourage" this. Because most of the time it just ends up being something we (as a table) have to fight against. I dislike "opinionated" systems--I prefer systems that provide tools but don't try to force one way of using them over another.

Hm, I mean something more like, the system gives people the option to 'escalate' to resolution at some kind of fixed, unavoidable cost or added risk. So basically, the people at the table are deciding 'is the outcome of this interesting enough to bite and spend the resource?' rather than having that decision originate from the GM.

I really don't tend to like what I call 'go-fish' game design in TTRPGs, where the GM is deciding on a set of gates that ask 'did you anticipate you would need to invest in X to Y level?' sorts of questions. So I like the idea of something that makes it so that players are going to be involved in the decision of whether something goes to a roll, and part of doing that is to make it so the end-member decisions of 'always roll' or 'never roll' are both strictly worse than actually making a determination based on context.

[PhoenixPhyre]

Hm, I mean something more like, the system gives people the option to 'escalate' to resolution at some kind of fixed, unavoidable cost or added risk. So basically, the people at the table are deciding 'is the outcome of this interesting enough to bite and spend the resource?' rather than having that decision originate from the GM.

I really don't tend to like what I call 'go-fish' game design in TTRPGs, where the GM is deciding on a set of gates that ask 'did you anticipate you would need to invest in X to Y level?' sorts of questions. So I like the idea of something that makes it so that players are going to be involved in the decision of whether something goes to a roll, and part of doing that is to make it so the end-member decisions of 'always roll' or 'never roll' are both strictly worse than actually making a determination based on context.

I ask for rolls...really really infrequently. At least outside of combat, where tiny little actions with built-in uncertainty are the norm. Like...generally a couple times per session per character. Because if what you're asking to do makes any kind of sense with the narrative so far (ie established fictional facts) and the character there just isn't all that much room left for interesting uncertainty. For social stuff, for instance, you only need to roll at the point you're asking someone to do something they're not willing to do given the reasons/motivations given, but could be convinced to do. And generally that's a narrow window. Most of the time it's obvious from the tenor of the conversation up to that point[1] which way it's going to go. Knowledge-type checks tend to be more about how much you know, not whether you know (ie degrees of success). And even those are on a sliding scale--it's not "DC X to know Y", it's "higher is better, but the baseline is different depending on who you are and what your character would know." The ranger with Favored Enemy: Elementals is asking about a blowing snow field in a vision after touching an elementally-charged scroll? Yeah, he's just going to know that it was of a place in the Plane of Ice. Which place, and other details, may require a roll. Or may not even be possible with what they know at that point. If the nerdy, but non-elementally-specialized wizard asked the same question (based on the ranger's description), he'd have to roll to distinguish it from a natural snow field. If the wizard's player tried to ask the question without the ranger describing it (ie meta), he'd get a sideways look like "dude, what? You don't know about that."

So...yeah. I guess I don't rely on "skill checks" (in particular) that much. So it rarely matters--cases where it does are generally obvious to everyone that the resolution should be done explicitly. In cases of ambiguity, I tend to lean toward the side of just not asking for rolls and assuming success if it doesn't really matter much. I prefer characters to feel competent.

[1] which may be 1st person, 3rd person, or summarized. Often it's summarized, because otherwise things get really verbose. And ain't no one (at least I don't) have patience for that.

[NichG]

I ask for rolls...really really infrequently. At least outside of combat, where tiny little actions with built-in uncertainty are the norm. Like...generally a couple times per session per character. Because if what you're asking to do makes any kind of sense with the narrative so far (ie established fictional facts) and the character there just isn't all that much room left for interesting uncertainty. For social stuff, for instance, you only need to roll at the point you're asking someone to do something they're not willing to do given the reasons/motivations given, but could be convinced to do. And generally that's a narrow window. Most of the time it's obvious from the tenor of the conversation up to that point[1] which way it's going to go. Knowledge-type checks tend to be more about how much you know, not whether you know (ie degrees of success). And even those are on a sliding scale--it's not "DC X to know Y", it's "higher is better, but the baseline is different depending on who you are and what your character would know." The ranger with Favored Enemy: Elementals is asking about a blowing snow field in a vision after touching an elementally-charged scroll? Yeah, he's just going to know that it was of a place in the Plane of Ice. Which place, and other details, may require a roll. Or may not even be possible with what they know at that point. If the nerdy, but non-elementally-specialized wizard asked the same question (based on the ranger's description), he'd have to roll to distinguish it from a natural snow field. If the wizard's player tried to ask the question without the ranger describing it (ie meta), he'd get a sideways look like "dude, what? You don't know about that."

So...yeah. I guess I don't rely on "skill checks" (in particular) that much. So it rarely matters--cases where it does are generally obvious to everyone that the resolution should be done explicitly. In cases of ambiguity, I tend to lean toward the side of just not asking for rolls and assuming success if it doesn't really matter much. I prefer characters to feel competent.

[1] which may be 1st person, 3rd person, or summarized. Often it's summarized, because otherwise things get really verbose. And ain't no one (at least I don't) have patience for that.

Well, I also want an investment in something like a skill to translate to a player being able to think 'by investing in this, now I can do this thing' or 'I'm investing in this in order to be able to do this thing'. That doesn't have to mean rolling for something - it could be some static threshold like 'you need 8 points in Arcane Lore and 3 in Pyromancy in order to use mental control over summoned arcane fire rather than pre-determining its behavior'. But for me it does mean that it's not good if the GM is the one who decides if and which skill is relevant to something. So having some way to give agency over that decision calculus to the player is good.

That sort of static threshold idea also works better with fewer gradations of things, so its easy to cluster related but not explicitly stated things someone might want to do. E.g. 'okay, you want to give mental control of your summoned flames to someone else? You want to change the color and make them speak words in your voice? Well Pyro 3 is 'controllable fire', and those are all examples of controlling fire, so if you also have Mind Magic 1 to make mental contact with someone else you're good to go for the first, and the second you can just do'

[gomipile]

Not a Battletech issue. Not comparing to Battletech. Chocolate is yummy, because sweet and chocolate. Dirt is yucky, because gritty and dirt. Giving examples of good and bad, not comparing the subjects of the comparisons.

Perhaps would have been clearer with the inclusion of phrases like “for example”.

That doesn't answer what I was asking you. To be more precise, in what RPGs do “more skilled characters fumble more often” to your knowledge?

Or is that something you just hope you never see happen, but don't have any examples of?

[PhoenixPhyre]

That doesn't answer what I was asking you. To be more precise, in what RPGs do “more skilled characters fumble more often” to your knowledge?

Or is that something you just hope you never see happen, but don't have any examples of?

If you take it straight and use the "nat 1 is a fumble" variant, 3e D&D does exactly that. Martials roll more attacks as they level up, so the chances of fumbling go up with levels, not down.

Yeah, that's kinda dumb.

[Lucas Yew]

Bell curve. Going further, I'd prefer all dice in game were the classic 6 sided ones (= d6); all others are not readily available where I live, but d6s are often usually found in local stationary stores.

[GloatingSwine]

If you take it straight and use the "nat 1 is a fumble" variant, 3e D&D does exactly that. Martials roll more attacks as they level up, so the chances of fumbling go up with levels, not down.

Yeah, that's kinda dumb.

"If you change the rules to be dumb, the rules become dumb"

Fumble on 1 wasn't part of the actual rules, I don't even recall it being a published variant, it was just a common dumb thing people did (like all those house rules for Monopoly that make it even more interminable and hideous to play that everyone seems to use).


It might be interesting to design around 3D6 but using advantage/disadvantage instead of arithmetic modifiers. Flat arithmetic modifiers on a bell curve are more valuable the closer to 50/50 the original check was going to be. 3D6+1 makes you 12.5% more likely to hit 11 but only 5% more likely to hit 16 for instance.

Whereas shifting to 4D6 drop lowest on an advantage roll shifts the 50/50 split to 13 from 11 and is still offering about double the chance of getting scores all the way up to 18.

Plus it's tactile, you pick up extra dice and then take one away (either the good one or the bad one depending).


You'd need to do a little more fiddling, but having stages of advantage/disadvantage instead of almost all of the modifiers where you add 1 dice and drop 1 either high or low would give you a system where you get a predictably shiftable bell curve. (I'd say maybe go up to 3 stages, so at most you roll 6D6 and keep the highest or lowest 3 before things get fiddly. You could maybe have any instance of boxcars or snakeeyes in the final result represent your crit/automiss if you want)

But I think you'd want to do it in more of a horizontal system than D&D which isn't designed to add ever bigger numbers to the same dice as you grow a character.

[Quertus]

Yeah. Ideally the way the system is structured would make this somehow obvious as to how to determine if consequences are interesting. This is sort of why I like 'roll to see how much something costs' rather than 'roll to see if you succeed' as a design element, though it didn't quite fit in this scheme...

Maybe something like, the most common way to get pips is to spend a relatively slowly-recovering resource, which can be done after the roll (even, say, hours after the roll), and a failed roll really means in all cases 'this is the price of that resource for you to do this thing'. So in no case can you ever 'try again' on any roll in the d20 sense - it costs what it costs.

I know people who don't like 'roll to determine things about the world' would hate that though.

I’m not sure. It sounds like you’re saying the equivalent of, “you can snatch the paper from the fire, but it’ll cost you a 2nd degree burn - are you willing to pay that cost?”. And I’m fine with that. Not so much with, “you failed to put on your pants, so you must be drunk” style of “roll to determine the state of the world”.

Hm, I mean something more like, the system gives people the option to 'escalate' to resolution at some kind of fixed, unavoidable cost or added risk. So basically, the people at the table are deciding 'is the outcome of this interesting enough to bite and spend the resource?' rather than having that decision originate from the GM.

I really don't tend to like what I call 'go-fish' game design in TTRPGs, where the GM is deciding on a set of gates that ask 'did you anticipate you would need to invest in X to Y level?' sorts of questions. So I like the idea of something that makes it so that players are going to be involved in the decision of whether something goes to a roll, and part of doing that is to make it so the end-member decisions of 'always roll' or 'never roll' are both strictly worse than actually making a determination based on context.

True, with “added risk”, it needn’t be “you *will* get a burn”, simply that, at your skill level, you *risk* getting such a burn *if* you commit to getting the paper. That’s arguably even more interesting, IMO.

I’m not sure how I’d most like to juggle what you risk by attempting real-time control of an elemental when you lack the optimal level of skills to accomplish that feat, however.

That is, gatekeeping “you can” behind a certain level of skill seems less interesting than “you unequivocally can at this level of skill; how do you compensate / what do you risk for each point of skill you lack?”. I’m just drawing a blank for any but the most cliché of answers to that question.

That doesn't answer what I was asking you. To be more precise, in what RPGs do “more skilled characters fumble more often” to your knowledge?

Or is that something you just hope you never see happen, but don't have any examples of?

Ah, well…

If you take it straight and use the "nat 1 is a fumble" variant, 3e D&D does exactly that. Martials roll more attacks as they level up, so the chances of fumbling go up with levels, not down.

Yeah, that's kinda dumb.

This is one of the common examples. To generalize, “if more skill produces more rolls, but an equal chance to fumble with each roll, that’s dumb”.

Another common example is the WoD-style dice pool / fumble system, where, at certain parts of the curve, more skilled characters are more likely to fumble than their lower-skilled counterparts.

Point is, if you make this class of mistake, you have failed to design a game component that I will find acceptable.

"If you change the rules to be dumb, the rules become dumb"

On a related note, people are dumb. This is simply a (sadly necessary) test to let them catch their mistakes.

[Cluedrew]

Determining if something is interesting is a quintessentially fact-and-table-bound problem. [...] I dislike "opinionated" systems--I prefer systems that provide tools but don't try to force one way of using them over another.

It can also be a campaign bound problem, which is why I actually do enjoy specialty system that are focused on a very particular type of game. Or I can, if they have the mechanics line up with that specialty. And I enjoy it a lot more than D&D trying to be everything to everyone.

Which is all preamble to say, bell-curve, although I can't back that up in any objective/universal way. I think it works better in environments without much scaling, but I'm not really into scaling anyways, and I can make similar comments about many of its other features. Although it might actually be the fact the size of the die involved tend to be smaller. I think a d20 is just to much variance unless all your modifiers are huge and if all your modifiers are huge why are we... OK there is an aesthetic in lots of big numbers, but I'm not really into it.

"If you change the rules to be dumb, the rules become dumb"

Oh boy, this is what sparked the discussion in the D&D-isms thread about how not all D&D-isms are actually rooted in the rules of D&D. I don't know how some of these rules got so popular, but it still serves as an excellent example of a bad rule. It (at least the version I talked my group out of) introduces negative consequences for actions in a way that is not affected by how skilled the character is, except when it is inversely affected by that. It also targets certain type of actions effectively randomly.

[NichG]

I’m not sure. It sounds like you’re saying the equivalent of, “you can snatch the paper from the fire, but it’ll cost you a 2nd degree burn - are you willing to pay that cost?”. And I’m fine with that. Not so much with, “you failed to put on your pants, so you must be drunk” style of “roll to determine the state of the world”.

Well, more like, lets say there's a castle wall you want to climb, which is scale 10 and you're scale 8 at climbing, so you need to roll 16 or higher on a d20 to climb it. Lets say you try to climb it and get a 3. Rather than the roll answering the question 'do you succeed or fail this one time?', this roll instead says 'climbing this wall costs you 13 points of stamina now and forever; if your scale changed by +1, it would bring it down to 10 points; if you got an extra pip, it would bring it down to 12 points, etc'. Of course if someone else tries to climb it, they get their own roll and own cost.

So its as if the 'unobserved' DC of the wall had been Scale 10(+0) e.g. 6 points above average for a scale 8 person, but because you rolled poorly then for you it determines the permanent 'observed' DC of the wall to be Scale 10(+7) e.g. 13 points above average instead.

In a videogame sense, its as if for every character, every obstacle or task or thing that could have a DC has a fixed random seed associated with it, and whenever you actually commit to doing the action with that obstacle you get to see what your random seed was. But if you save/reload or try the tasks in different orders, you can't trick the RNG because the seed is fixed and immutable per task rather than just drawing the next number off of the stack. The sort of tricky thing though is that it means that if you had elected not to roll and just take 10 instead, the stamina cost to just climb the wall would have been lower than what you ended up with due to a bad roll. So in some sense, the wall does have this quantum aspect that if you are in a rush and try to force it with a roll rather than taking your time it could somehow look like that 'changes the nature of the wall' for you. You could get around this by basically saying 'well, if you approach the wall with all the time in the world to take 10 some time in the future, you can use the wall's base scale rather than your previous result to determine the cost of success' - e.g. 'approaching the wall carefully' and 'approaching the wall in a hurry' are just different tasks, and whatever that extra 7 cost was about from the roll it was something that only matters if you're rushed or under pressure.

True, with “added risk”, it needn’t be “you *will* get a burn”, simply that, at your skill level, you *risk* getting such a burn *if* you commit to getting the paper. That’s arguably even more interesting, IMO.

I’m not sure how I’d most like to juggle what you risk by attempting real-time control of an elemental when you lack the optimal level of skills to accomplish that feat, however.

That is, gatekeeping “you can” behind a certain level of skill seems less interesting than “you unequivocally can at this level of skill; how do you compensate / what do you risk for each point of skill you lack?”. I’m just drawing a blank for any but the most cliché of answers to that question.

That's where something like resource pools could come in. For this example, lets imagine that a character has a 'stamina pool' that lets they pay the costs for success when their flat level of skill wouldn't cut it. Furthermore whenever a player calls for a roll instead of e.g. 'taking 5' it always costs an extra 3 stamina paid up front (but if they don't choose to succeed at that point, but come back to it later, they don't have to pay that 3 extra stamina again). The system would be designed so no one can ever be forced to roll by something external to themselves - you can always take 10 in calm situations where you can take your time or take 5 in situations where there's pressure and a cost to failure. So for that 3 stamina on average you get a +5 bonus in strenuous situations - a good incentive to roll in strenuous cases even if you dislike the variance. But at the same time, rolling when it does't matter would normally be silly because you're just throwing away stamina. If stamina regenerates at, say, 1 point per hour per Body scale and you have a total pool equal to 4*Body stamina, calling for a roll isn't ruinous but there is a bit of a bite if you do it too wastefully outside of downtime situations.

Now lets take 'putting on your pants' for example. Well, base characters just auto-succeed even taking 5. What if someone intentionally makes it really hard on themselves somehow though? The narrative cost of 'trying to put on your pants by throwing them in the air and doing a backflip so they land perfectly on your legs' might be tricky to make actually meaningful, especially if e.g. you're doing this where no one is around to see it, in a space with sufficient padding, etc. One valid way to resolve that is to just say 'yeah that's hard enough you have a chance of failure, but the result doesn't matter because you took a bunch of precautions, so you try it a bunch of times and succeed only some of the time but we won't roll for it, lets just move on'. Which, fair enough!

But at least in this design hypothetical, lets say it really matters to the player to get an answer about mechanically 'what is it like to do this trick?'. If we go by the first paragraph of this post, where rolling against an 'unobserved' thing sets the cost in the future, you could imagine a situation like a player wants to develop a line of stunts for a circus show, they have a stamina pool of say 16, and they want to roll in advance so they can figure out which tricks they can chain together and still afford the stamina to do the entire sequence consistently. In that case, rolling to put on their pants (in this ridiculously over-difficult way) could actually be made meaningful by the player's intent with what they're going to do with that going forward. So they say 'yeah I know this doesn't matter now, but I want to roll'. The GM says 'okay, pay 3 stamina now, and what you get now will determine what it will cost you during your performance later'. E.g. you can pay 3 stamina to 'lock in' the cost of a stunt that you intend to repeat later on. That makes the roll significant (because if you roll a 1, that backflip pants-mount trick will forever be one of the more costly things in your repertoire, and if you roll a 20 it will forever be one of your easier tricks that is basically free to execute), and the act of calling for the roll had a cost which lightly disincentivizes calling for it when there's no actual meaningful stakes, but light enough that its still an easy option for a player to choose most of the time.

And of course you could make things a bit more flexible and say that when you 'roll to establish cost', those costs aren't permanent and forever, but persist for say, 1 day or 1 week. Enough to prep for your circus act, not enough that a player will be incentivized to spend hours trying every single stunt they can think of in advance so they can lock in the prices. These examples are more to give an idea than ruleset final form stuff...

[GloatingSwine]

I think you'd want to design that system the other way around.

Design the costs and resource pools first, then figure out how characters will use skills, risks, and alternate resources to reduce those costs because the costs become the limiting factor on how much and what you can can do faster than the other things.

[NichG]

I think you'd want to design that system the other way around.

Design the costs and resource pools first, then figure out how characters will use skills, risks, and alternate resources to reduce those costs because the costs become the limiting factor on how much and what you can can do faster than the other things.

Well, the spirit of this thread is basically 'what's a good form of random variation?'. The principle then is, in turn, what questions whose answers I don't know do I want to ask? And what I want to do with the idea is to specifically not have the dice be a vehicle for the GM to ask the player 'does your character succeed?'.

So the kind of keystone of the construction is the idea that the dice should be asking some other question, and what kinds of questions might work there. That in turn suggests possibilities like 'ask the dice how much something costs', and then that conceit is what creates the need for the system to have things like fungible costs, which in turn is answered by design elements like resource pools, and so on and so on.

[firelistener]

Personally, I prefer a bell curve, perhaps because I'm a bit risk-averse. It's why I usually wield a greatsword (2d6) instead of a greataxe (1d12) when playing a barbarian in D&D. Even if it doesn't have the massive damage potential, I like avoiding low rolls more than I care about getting the high rolls. I think it helps me predict what my character is capable of doing.

[martixy]

For 3.5e and the game I usually want to play - flat.

The numbers are tuned way too deeply for any changes, but I'd probably do 2d20 if I could due to the power creep inherent in that system. I'm a fan of shallow bell curves.
It also makes advantage mechanics more interesting - from roll 2, pick the best/worst, to roll 3, drop the lowest/highest and even dice pool things.

[Quertus]

what questions whose answers I don't know do I want to ask?

What questions do I want answered? Combat as War questions: what amount of bonuses or penalties do I need to stack in order to not have to roll? That’s my primary question.

its as if for every character, every obstacle or task or thing that could have a DC has a fixed random seed associated with it,

Sounds good to me.

[KorvinStarmast]

Chainmail used 2d6 dice to resolve various conflicts.
The original D&D game used 2d6 for turning undead chances and 2d6 for loyalty and morale checks.
It also used an alternate combat system with a d20 based on the desire to apply the armor class idea (which had some roots in Don't Give Up The Ship) with something other than 2d6 as the basis.

A wide variety of tables had percentile dice generate stuff, and various d6, d8, 10 and d12 for random generation of stuff for dungeon encounters.

Game was still fun. Going to a 'unified d20 basis' was a decision WotC made for what I think were aesthetic reasons, and in an attempt to stream line the game a bit.

[LibraryOgre]

It really depends on the "feel" of the system. The idea behind D&D combat is "the chaos of combat", anything can happen, etc. How well-executed that is can be argued, but that's the intent.

More bell-curvey systems, like d6 systems, typically have an element of "cinematicness" baked into them. They are largely a power fantasy where skill plays more of a role in outcomes than luck.

Both are fun, but one type of rolling doesn't really work for a game trying to evoke the opposite feel.

I was thinking a bit differently.

IME, linear die systems tend to be associated with level-based systems, which also generally assume a fairly steady and somewhat uniform increase in ability.... while the fighter may become a better fighter, and the wizard a better wizard, both get improvements in a variety of things as they increase in level. IME, this makes for a more "cinematic" game, with characters pulling off cool stuff that they absolutely could not a few levels ago.

Distributive systems tend to be in skill based systems, where allocation of experience-equivalent might make someone a great fighter, or a generalist, or turn a fighter into a wizard, or what have you. This makes calibration from a GM a bit more difficult... even two games where the PCs have both earned X experience can be wildly different, depending on the choices of the player.

Distributive systems tend to emphasize middle-of-the-road results... truly great exploits aren't common. Linear systems are more "swingy". If the roll is 3d6, and massive success is on 18, that's a 0.5% in a distributive system, while the chance of a 20 on a d20 is 5%... ten times as often. To reach that frequency in a distributive system, 16-18 needs to be the critical success range. If the system allows for catastrophic failures (fumbles), then you have those great lows, too, as more frequent in a linear system.