How the DM is in theory meant to run ability checks:
1.set a DC
2. if the player rolls over it, they succeed, otherwise not
How I think everyone actually does it
1. don't set a DC
2. player rolls
3. about 10 or less is a failure
4. about 19 or more is success
5. 11-18, the DM does the thing that makes the story more interesting

or am I unusual in this?

@sil I usually don't set the DC first either. I prefer to see the roll first, and then make a ruling. Although my DC goes from 8 to 30 depending on how easy/hard it is, and I do try to keep it objective.


So a couple of problems with the “setting difficulty after”…

  1. it makes the actual stat worthless and makes the printed number on the die everything. This is where the “d20 is so swingy” myth cames from, if you banana peel moment on a nat 4 or lower and heavens part on a 17+ and your printed bonus, expertise or w/e is never looked at.

  2. it’s so easy to cheat. I never fudge now but I sometimes say that I used to fudge all the time (not woth current group). But it’s not exactly that I would change die rolls. It’s more that I used to do the “setting difficulty after” thing. It quickly became “stat doesn’t matter” and then it devolved into “roll doesn’t matter, either, I’ll just say what I want regardless of the roll”

  3. It’s slower. I know, I know, “Mercer does it” but “Make a strength check, DC 15” is faster than “Make a strength check.” “12.” “Ok, you fail.” and costs less of your attention. It’s also easier math for them; they have +7 they know that only nat eights or higher succeed. This is is especially if there are gonna be several rolls against the same number, such as DC vs an AC.

  4. There is also tension value in them knowing exactly what they need to roll, especially when there are high stakes. If you like me, roll over six, if you don’t, I’ll die. Les jeux sont faits. Rien ne va plus.

  5. Arguably setting after is itself alrrady cheating. It breaks “No paper after seeing rock” principle.

So I always say AC or DC as I ask for a roll.

Most of the time the DC is given from the NPC they’re interacting with, from the module/prep, or from a rule in the DMG. Almost always.

But when it’s not, I have a couple of standards.

  • 40 for “stfu or gtfo” asks like “Do I know the quaint and curious bit of forbidden lore that empires have been shattered to bury?”
  • “The one percenter” when something is so easy, but high stakes, there’s only one percent chance (DC 2 with unmodified dice but advantage) that the bad thing will happen (carrying a valuable egg up a ladder —- even non-adventurers can climb ladders, buuut that egg looks mighty precious), or, conversely, it’s almost a hard no but maaaaybe fate will smile on them (DC 19, disad, unmodified dice).

And then my go-tos. How many outcomes do immediately come to mind? Two? (“Pass/fail”, “live/die”…) then DC 15 (on the modified roll ofc, not natural).

More than two? I call this a twelvetwenty and it might be interesting to you, @sil. Eleven or lower, they get the worst outcome. Twenty or higher, they get the best. Twelve, but not twenty? They get their pick among the rest if there is more than one mediocre outcome, or the one mediocre outcome if I can think of only one. I try to state these outcomes outloud before the dice are rolled. Because murky DM authoity is Not Cool while clear stakes are So Awesome.♥


@Sandra @sil (part 1)
Great points as always, Sandra!
I do need to clarify that players tell me the numbers after adding the modifiers(I don't see the rolls), and if there is a high-stakes roll, I'll say the DC up front. That really cranks up the tension.
If you don't say the DC up front, players won't know if they've succeeded or not. This can be used to create suspense and tension, especially in rolls with scope for reveal through narration.

@pench4nt @sil In my experience it often causes apathy, or that’s how I feel at least, whwn playing. It feels like the roll doesn’t matter, or is just theater.

@Sandra @sil
That's fair. And I don't know what happens if all the DCs are revealed up front. I think I'll run a few games where I do that and see if it changes anything for the players :)

@Sandra @sil
(part 2)
One place I will disagree with you is the swinginess of d20.
d20 is swingy because all the numbers are equally likely to come up. So, a +5 modifier has 5 times the probability of a +1 modifier (25% vs 5%). In a less swingy system, like 3d6, a +5 modifier vs +1 is 45% vs 13%.
The modifiers are much more valuable in a less swingy system compared to a swingy one. Not saying one better than the other, but d20 is more swingy than 3d6 or 2d10.

@Sandra @sil
The once percenter is super interesting, I think I'll steal that :P

I like the blorb principles too.

@Sandra @sil
We need to define what one means by "swinginess" before attempting to answer if 1d20 or 3d6 is more swingy.
I would define swinginess the as extent of variation in results. More swingy means more variation, and less swingy means less varied, more predictable results.
Would you agree?
Also, the character limit is preventing me from properly explaining my point of view 😅​
If you are interested in continuing this conversation, I'll write it out elsewhere else and share it.

We both know that on 3d6 a ten is more likely than a three♥


We need to define what one means by “swinginess” before attempting to answer if 1d20 or 3d6 is more swingy.

That’s my point; people having been conflating various meanings of swinginess. (Sort of like how people use “granular” to mean both more fine-grained, and it’s opposite, more coarse-grained.)

This leads to some misconceptions. I saw a guy on a blog wrongly claim that 1d6+3 was more swingy than 2d6 “because more dice is always less swingy than one linear die” 📊

Sure, an encounter table that is on 2d6 will have those middle results (yer basic wolves and what not) more often than the outliers, there is a more rare one-in-thirtysix chance that a dragon will show up.

So that’s when you want multiple dice—when you want a chance for extreme results.

I’ve seen many people online, not including you among them ofc, who are like “+5 vs DC 15 on d20 is more swingy than +5 vs DC 15 on 3d6”. Because they just parrot the “more dice = more swingy” thing.

Well, when DMs make the outliers especially significant, and linear doesn’t really have outliers, every result is equally likely… then yeah, d20 is gonna feel more swingy.

I would define swinginess the as extent of variation in results.

But when you’re doing binary pass/fail, there is no variation in result. There is X chance of success and 1-X chance of failure. You just have DMs, module writers, and game designers have a more intuitive understanding of what those chances are. And some games come around with completely broken understanding of probability (like City of Mists, GURPS, the Burning Wheel, or the original 90s version of World of Darkness) while some games can handle multiple dice just fine (like the original AW, or Fudge).

If you define swinginess as more variation in the results, then with multiple dice you get less swingy in the middle and more swingy at the edges.

If you define swinginess as less variation in the results, then with multiple dice you get more swingy in the middle and less swingy at the edges. Arguably this is where we’re at because modifiers are so sensitive in the middle. A +1 and you’re suddenly twice as likely to hit, that kinda thing… (I get that the point is to get diminishing returns as the mods pile on but the games made on that model just so often work very poorly in practice.)

Meanwhile linear just matches our everyday intuition of results. One in six, one in a million, one in twenty, 20% cooler etc.

So yeah, when you do want a chance of outliers in variation, use curves/pools. Perfect for encounter tables & damage rolls. When you don’t, use linear. Especially when you’re using modifiers or moving target numbers♥

(part 1) I agree on the point that dice pools are useful when you want the possibility of extreme results. Another situation is where you'd want a large difference between the abilities of PCs, without making something impossible. This is possible to be achieved by d20 too, but here is a difference. I'm finding it difficult to articulate right now. I'll try and share that later :)
Their intuitiveness is a big plus too.

@Sandra (part 2) I've run Mage The Ascension, which has a dice pool system. I do like the mage system, but it takes some getting used to. The probabilities aren't as straightforward as a uniform distribution like d20.


The original Mage had greater chance of botching the better you got. Dice pools are just difficult for most devs and designers to work with unless they really know what they’re doing.

Another situation is where you’d want a large difference between the abilities of PCs, without making something impossible.

Ah, not really, since pools clump outcomes into bands giving you fewer possible results among the possible outcomes.

1d100 = A hundred different results, i.e. stat levels to choose between. 2d10 = Only 19 different results (from the hundred outcomes).

I don’t mean variation of results with one roll, but variation in output across many rolls.

Aha. There’s no difference between curve vs linear in that regard.

2d10 and 1d100 has the same chance of rolling snake eyes♥

3d6 eats its own tail; sure, it makes the outliers (3, 4, 17, 18) less likely but it also makes the common numbers (10, 11) more likely. But it gets super volatile there in the middle: The step from 10 to 11, is 12.5 percentage points!

Eh, you know all this stuff. We’re arguing semantics at this point.

Dice pools are indeed difficult to work with, unless one has a good theoretical/practical understanding of probability theory. I think that mage has done it reasonable well.
And yeah, it's just about the semantics at this point 😅​


What I mean is for the first seven years of Mage’s lifespan (the first two editions), you had a higher chance of botching the bigger pool you had. This was fixed in revised edition (basically third edition, 2000).

@Sandra Aah, I see. That intuitively makes sense to me, because better mages are less likely to screw up a spell. But intuition is often wrong.
It does a long time to figure out and fix such things, eh? :P

(part 3) I don't mean variation of results with one roll, but variation in output across many rolls. My idea of swinginess is something akin to standard deviation. i.e, the likelihood of extreme results to occur in a chance event (die roll). This will also have to be normalized to 1 because otherwise, large values will give a large standard deviation. By this definition, d100 and d20 are equally swingy, 2d10 is less swingy than that, and 3d6 is even less swingy.

Sign in to participate in the conversation
Tabletop Social

We are an inclusive Mastodon community for everything tabletop (and more).