Bell curve versus flat die roll preferences

[White Blade]

When you’re playing a TTRPG do you generally prefer a bell curve dice roll where the results tend to congregate around the middle or do you enjoy the swingy rolls of a d20 system?

[KorvinStarmast]

When you’re playing a TTRPG do you generally prefer a bell curve dice roll where the results tend to congregate around the middle or do you enjoy the swingy rolls of a d20 system?

Both. It need not be either / or.

[Rynjin]

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.

[Satinavian]

bell curve in nearly every instance.

[animorte]

I've always liked the idea of the bell curve, but only used it a little bit. It seems the intent is to standardize the outcomes. I actually think that it's better to provide a more clear system of being really good at what you're actually good at while still leaving the door open enough for those highs and lows.

[Mastikator]

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 100% agree with this.

I will however add that flat die results are less compatible with degrees of success/failure. I think especially critical failures and fumble tables only really belong in games that are supposed to feel cartoonish and silly, like Toon.

[False God]

bell curve in nearly every instance.

Agreed.

----

Personally, I've replaced the d20 in my D&D games with 2d10.

[Xervous]

Sorta bell curve but not quite? Yes it’s curved, but most importantly the curve widens and warps rather than just marching up so many pegs.

[Easy e]

Bell Curve all the way.

I prefer dice pool systems, or roll and Keep methods.

[Telok]

Flat for comedy & games where lol-random or chaos & unpredictability should reign. Bell curves or some variation thereof for anything else.

[johnbragg]

What I keep tinkering with is a character class built on the ability to switch between them. Inspired by Haley's speech about Xykon not being any of the three visible Xykons at Azure City -- the rogue cheats at solitaire, the rogue cheats at reality.

So I want a sneaky type who has the ability to choose to roll 3d6 ("Play It Safe") or d20 ("Swing For The Fences") as a class feature.

[Quertus]

For me, that’s the wrong question, so… mu?

The resolution method should match what the system is going for, and what I expect from the system.

The resolution method should *always* be applicable. What is the DC to put on my pants? What are the stats for a small child? What are the penalties for being drunk? We shouldn’t *have* to break to GM narration to get reasonable results from attempting to put on pants.

What do you know about the history of the cult of Bane in 2e FR? What do I know? If the system has “knowledge” skills, it had better be able to model that difference.

What does the end result look like when you write code, or a short story, or a post? What does the end result look like when I do so? Does the system fight you to model that difference?

The 2d6 bell curve of Battletech is very well suited to making hunting for those “+1”s meaningful. Contrariwise, “more skilled characters fumble more often” is antithetical to my desires.

But, really, I prefer “we’ve stacked the deck so hard, we don’t *need* to roll” as the default resolution method. Combat as War, and all that.

So… whichever resolution method makes it feel appropriate that we’re actually touching the dice? 3e bonuses outgrowing Arangee is preferable to “bounded accuracy”, for instance.

Lastly, as I prefer the granularity of well-implemented “degrees of success”, a flat d20 or a dice pool of successes or a dice pool of summed results are preferable to normal bell curve implementations.

So, to finally answer the question asked, if push comes to shove, I prefer a flat d20, as it has a higher ceiling on implementation fidelity than any bell curve system I’ve seen.

[stoutstien]

Hmm I like a hybrid of both plus a little low roll protection built in for those who are "proficient" in that subset.

[stoutstien]

What I keep tinkering with is a character class built on the ability to switch between them. Inspired by Haley's speech about Xykon not being any of the three visible Xykons at Azure City -- the rogue cheats at solitaire, the rogue cheats at reality.

So I want a sneaky type who has the ability to choose to roll 3d6 ("Play It Safe") or d20 ("Swing For The Fences") as a class feature.

In my homebrew system that is something I'm toying with. Not only that but depending on "how" they choose to attack (power or precision) the floor/ceiling and minimal damage will defer. So a sneaky type might rely on a well placed strike and if it hits it hits hard where the more fighter type might make strikes that are less opportunistic but wear down the target regardless.

[Pauly]

Short answer, it depends on what the roll is for. For simple routine tasks a prefer a flat roll with no modifiers. For complex tasks I prefer a 3dx system.

Long answer, in a well designed system it doesn’t matter.

The fly in the ointment is more modifiers and how they’re implemented rather than what you’re rolling.

[stoutstien]

Short answer, it depends on what the roll is for. For simple routine tasks a prefer a flat roll with no modifiers. For complex tasks I prefer a 3dx system.

Long answer, in a well designed system it doesn’t matter.

The fly in the ointment is more modifiers and how they’re implemented rather than what you’re rolling.

1 max IMO. Same for any type of reroll options.

[ToranIronfinder]

Dice pool with a wild die, it's bellcurvy, so abysmal failures and swinging for the fences are both less likely to succeed, but the wild die gives you some uncertainty.

[PhoenixPhyre]

I care much more about simplicity of resolution compared to curved/flat.

Flat with a crap-ton of modifiers, reroll abilities, opposed rolls, etc? No thanks. Similarly when simple operations need a bunch of conditional steps or a flowchart to resolve. Or when similar actions require radically different resolution mechanics (looking at you AD&D with the combination of d20 + mods roll over, d20 roll under, % roll under, sometimes adding, sometimes subtracting, sometimes high is good, sometimes low is good...)

Flat with simple, limited modifiers and not much other jank? Fine by me.
Curved with simple, limited modifiers and not much other jank? Fine by me.

Curved with a bunch of complexities? No thanks. It's one reason I'm not so fond of dice pools or counting successes or "unusual" dice or exploding dice. I don't want to have to remember which die is which and want to be able to roll a bunch of things at once. As a GM, if I have to roll 20 attack rolls with the same modifiers...I want to just be able to roll 20 dice and allocate appropriately.

[gbaji]

It kinda depends on where your success/failure probabilty range should lie. I do somewhat like flat systems, especially if there's a scrolling range involved (which D&D sorta has). it works. It presents a consistent set of probabilities within a given range (plus or minus X numbers off a "even" chance).

Bell curves have their benefits (more consistent results), but only if the success/fail range is somewhat (or entirely) static. Meaning that you succeed on X or higher/lower, and counter factors don't matter (or don't matter much). Where bell curve have problems is when there's some sort of scolling difficulty range involved.

Example: I'm trying to hide and you're trying to spot me. It's an opposed skill check. Let's say I have a skill level 18 on my hide, and you have a 15 on your spot. We can normalize that on a D20 by saying that 1-10 fails, and 11+ succeeds, with the levels being plusses or minuses, meaning that I'm 3 levels higher, so if I roll a 8 or higher, I succeed. Simple, and scales automatically to any relative skill levels (even to ones where the difference is sufficient that auto-success/failure occurs).

Same thing on a bell curve results in significantly greater differences in overall probability. A few points of difference has a much larger resulting probabilty change. It makes opposing skill checks "wonky", since at different deltas, the actual value of a +1 changes. Which seems "odd".

If we're doing a just a flat (rolll 3d6 against your skill), instead of opposed checks, then we see that a small number of bonuses have a very large effect initially, but a set of diminishing returns as we increase them. This can have an advantage in a game of encouraging people to spead out skill points instead of just increasing them (a problem in scalable flat die systems). And that can be a good thing.

But yeah, the cost is that on a bell curve there's a *very narrow* range of "it's a toss up or close" with the rest falling into "pretty much guaranteed success or failure". Which can have the effect of narrowing the difficulty range the game itself can manage. A slight bit easier, and your characters walk through the encounters, a slight bit harder and you get curbstomped.

I don't think there's any one "best" way to do things. It depends heavily on how your skill system works, and how rolls are resolved. I would be hesitant to just change a D20 into say 2d10 or 3d6 though, since the calculations surrouding the roll itself are based on assumptions about the probability changes based on incremental bonuses in the check itself. You really need to change more than just the dice if you want to do that IMO.

[stoutstien]

I care much more about simplicity of resolution compared to curved/flat.

Flat with a crap-ton of modifiers, reroll abilities, opposed rolls, etc? No thanks. Similarly when simple operations need a bunch of conditional steps or a flowchart to resolve. Or when similar actions require radically different resolution mechanics (looking at you AD&D with the combination of d20 + mods roll over, d20 roll under, % roll under, sometimes adding, sometimes subtracting, sometimes high is good, sometimes low is good...)

Flat with simple, limited modifiers and not much other jank? Fine by me.
Curved with simple, limited modifiers and not much other jank? Fine by me.

Curved with a bunch of complexities? No thanks. It's one reason I'm not so fond of dice pools or counting successes or "unusual" dice or exploding dice. I don't want to have to remember which die is which and want to be able to roll a bunch of things at once. As a GM, if I have to roll 20 attack rolls with the same modifiers...I want to just be able to roll 20 dice and allocate appropriately.

Aye. Been pondering over this for a long time and I realized that most unnecessary complexity is added in as distractions to simple flaws in design. The resolution mechanic either lacks engagement or lacks impact.

[PhoenixPhyre]

Aye. Been pondering over this for a long time and I realized that most unnecessary complexity is added in as distractions to simple flaws in design. The resolution mechanic either lacks engagement or lacks impact.

Personally, I want a resolution mechanic that stays out of my way and does only what it's asked to do, which is be a tunable coin flip between acceptable alternatives. I'm not playing the resolution mechanic, it's just there to help me move the important part (the fictional set of events) along in a satisfying way.

[animorte]

Aye. Been pondering over this for a long time and I realized that most unnecessary complexity is added in as distractions to simple flaws in design. The resolution mechanic either lacks engagement or lacks impact.

Yes, this dice concept comes up and it still goes back to all those times that you probably shouldn't need to roll the dice anyway.

[PhoenixPhyre]

Yes, this dice concept comes up and it still goes back to all those times that you probably shouldn't need to roll the dice anyway.

Strong agree. Most of the "bad results" of flat dice come from asking the dice to do too much of the thinking for you. I personally never ask the dice (or the rules generally) to answer any question where I'm not ok with any of the possible results. Either I'll prune the possible results tree or just not ask the dice to resolve that.

[stoutstien]

I'm playing with an ability test system with 2d6+mod. expertise/additional/features only increase minimum values one one die. no rerolling or fuss. With +3 being the soft cap on modifiers it's a wonderful little package.

[PhoenixPhyre]

I'm playing with an ability test system with 2d6+mod. expertise/additional/features only increase minimum values one one die. no rerolling or fuss. With +3 being the soft cap on modifiers it's a wonderful little package.

The (very WIP) system I'm working on (slowly) uses a d20 + fixed mod (based on approach, not task) + circumstance die vs a TN. All circumstantial modifiers are lumped into the circumstance die, which ranges from -1d10 to +1d12. Basically all the features that would impose penalties just slide that circumstance die value around. Training and expertise are done by setting floors on the result.

Modifiers are 0-5, and different classes have specific archetypes (ie approaches) that they're trained at. For instance the mage has fast progression with Knowing approaches (book learning, basically) and moderate progression with Precise approaches (detail-oriented, fine motor skills). The Brute class (name still WIP) has fast progression in Strong (brute force, direct head on) and moderate in Willful (mental endurance, mental toughness) approaches. You're guaranteed at least a +2/+1 in your primary/secondary archetypes from your class.

[gomipile]

The 2d6 bell curve of Battletech is very well suited to making hunting for those “+1”s meaningful. Contrariwise, “more skilled characters fumble more often” is antithetical to my desires.

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?

[Notafish]

I think that 2- and 3d6 bell curves are nice - easy to estimate, and the chance of a dice swinginess feeling like it is overwhelming mods due to skill feels lower, especially if you are playing a game where action scenes are narrated blow-by-blow. bell curves on d8s and up take just enough extra work for me to estimate that it would take some work for me to buy in to the reasoning. d2s and d4s would be a little awkward doing addition rather than a dice pool, imo.

With that said, I think the bigger dice are fun to roll. I think I would like a DnD-type game that used 3d6 for most things but kept the d20 for things like initiative and oracle rolls.

[ngilop]

The (very WIP) system I'm working on (slowly) uses a d20 + fixed mod (based on approach, not task) + circumstance die vs a TN. All circumstantial modifiers are lumped into the circumstance die, which ranges from -1d10 to +1d12. Basically all the features that would impose penalties just slide that circumstance die value around. Training and expertise are done by setting floors on the result.

Modifiers are 0-5, and different classes have specific archetypes (ie approaches) that they're trained at. For instance the mage has fast progression with Knowing approaches (book learning, basically) and moderate progression with Precise approaches (detail-oriented, fine motor skills). The Brute class (name still WIP) has fast progression in Strong (brute force, direct head on) and moderate in Willful (mental endurance, mental toughness) approaches. You're guaranteed at least a +2/+1 in your primary/secondary archetypes from your class.

Alternity's resolution mechanic is similar to this.

[NichG]

There's a system I want to go back to one of these days and figure out how to make it ergonomic: a flat roll that acts as a multiplier on an innate quantity.

So for example, if your base effect magnitude (from skill, level of the power, attribute, whatever) is 20, you roll say 1d10 and multiply the base by 100% + 10% × result - so a 7 becomes a 34, etc.

This has the advantage that it has a fixed relative variance regardless of the scale of the base effect. Works as well for 2 as 2 billion. So you can independently manipulate variance (by changing the die size or % per point) and effect scale. The problem is, it involves division, which is terrible.

Maybe just something like half on 1-2, equal on 3-8, double on 9-10?

[stoutstien]

The (very WIP) system I'm working on (slowly) uses a d20 + fixed mod (based on approach, not task) + circumstance die vs a TN. All circumstantial modifiers are lumped into the circumstance die, which ranges from -1d10 to +1d12. Basically all the features that would impose penalties just slide that circumstance die value around. Training and expertise are done by setting floors on the result.

Modifiers are 0-5, and different classes have specific archetypes (ie approaches) that they're trained at. For instance the mage has fast progression with Knowing approaches (book learning, basically) and moderate progression with Precise approaches (detail-oriented, fine motor skills). The Brute class (name still WIP) has fast progression in Strong (brute force, direct head on) and moderate in Willful (mental endurance, mental toughness) approaches. You're guaranteed at least a +2/+1 in your primary/secondary archetypes from your class.

Freaky. Great minds think alike. I use a die that replaces one of the D6s to represent advanced mastery in a field that also grows and shrinks with stuff like cover for sneaking. Mine never drops below the original d6 though.
As an alternative they can roll just a flat D20 as a fate check against the original test DC once per day. Some archetypes get some advantages on this but it never gets better than a coin flip.