Why Blades in the Dark feels less swingy than d20 – and why the bell curve (and variance) aren't the main reason

After recalculating the numbers, I believe the common explanation "d20 is swingy because uniform distribution vs. bell curve in dice pools" misses the real core difference. The bell curve and variance per roll are largely irrelevant here.The key lies in two other mechanisms:

1. Granularity of progression / how difficulty & competence are modeledIn d20 systems like PF2e the probability curve of a single roll stays exactly the same shape (shifted uniform).
   When your character gets better (+1 bonus) or the task gets easier (-1 DC), the entire curve shifts linearly by 5 %. Every improvement is fine-grained and perfectly incremental.In dice-pool systems like Blades in the Dark each fixed number of dice (1d6, 2d6, 3d6, 4d6) produces its own completely distinct, fixed probability distribution ("template").
   This template cannot be shifted.
   To make a roll easier or harder you switch to a different template (more or fewer dice).
   That means improvements happen in large, non-linear jumps (e.g. going from 2d6 to 3d6 halves failure chance from 25 % to 12.5 %, while partial success stays almost the same).Edge probabilities (worst & best outcome) for Blades (highest die decides):
   • 1d6: 50 % Failure / 16.67 % Success / 0 % Critical
   • 2d6: 25 % Failure / 27.78 % Success / 2.78 % Critical
   • 3d6: 12.5 % Failure / 34.72 % Success / 7.41 % Critical
   • 4d6: 6.25 % Failure / 38.58 % Success / 13.19 % Critical
   At 3–4d6 the edges (extreme failure and critical success) are already in the 6–13 % range – very similar to typical PF2e crit fail / crit success chances (~5 % each).
   This suggests d20 is not inherently swingier than dice pools when looking at extreme outcomes. The bell curve / variance explanation therefore doesn't hold up as the primary reason.
2. Rightward shift of outcome labels (Position & Effect)In Blades, Position (Controlled / Risky / Desperate) and Effect level systematically shift the meaning of the dice results to the right:
   • What would normally be a failure becomes a partial success
   • Partial success becomes full success
   • Full success becomes a critical
   As a result, true hard failure (scene-stopping catastrophe with no progress) shrinks to ~5–10 % or less (often even lower in good position).
   The game stops being about "do you succeed or fail?" and becomes almost exclusively "how well do you succeed?"In contrast, in most d20 scenarios a miss/failure stays in the 40–50 % range – there is no comparable systemic mechanism that eliminates hard failures on that scale.

Summary:
The feeling that Blades is "less swingy" or "more reliable" doesn't primarily come from bell curves, lower variance, or better edge probabilities.
It comes from the combination of
(a) chunky, non-linear competence jumps (switching fixed templates)
and
(b) active, mechanical minimization of catastrophic failure through rightward shifting of outcome labels.This seems to be the deeper design difference that actually explains the player experience – not the shape of the probability distribution on a single roll.What do you think? Am I missing something important here?

My lived experience suggests that you're not accounting for a statistically untenable propensity to roll 3 or less on each of any number of d6. Blades is a great game, but, boy howdy, I've had some garbage luck in it. It doesn't matter what dice I use in what combination, they all turn up rough. I'd roll higher if I rolled d3s.

franz101 said:

• 1d6: 50 % Failure / 16.67 % Success / 0 % Critical
• 2d6: 25 % Failure / 27.78 % Success / 2.78 % Critical
• 3d6: 12.5 % Failure / 34.72 % Success / 7.41 % Critical
• 4d6: 6.25 % Failure / 38.58 % Success / 13.19 % Critical

Click to expand...

why do your percentages never add up to 100, shouldn’t they? They do not even add up to the same total…

mamba said:

why do your percentages never add up to 100, shouldn’t they? They do not even add up to the same total…

Click to expand...

My bad.

• 1d6: 50 % Failure / 33,33 % Partial / 16,67 % Success / 0 % Critical
• 2d6: 25 % / 44,44 % / 27,78 % / 2,78 %
• 3d6: 12,5 % / 45,37 % / 34,72 % / 7,41 %
• 4d6: 6,25 % / 41,98 % / 38,58 % / 13,19 %

franz101 said:

The game stops being about "do you succeed or fail?" and becomes almost exclusively "how well do you succeed?"In contrast, in most d20 scenarios a miss/failure stays in the 40–50 % range – there is no comparable systemic mechanism that eliminates hard failures on that scale.

Click to expand...

That’s interesting because when I played Scum and Villany (also a FitD game), I had the same feeling but it felt more like everytime we wanted to do something, we failed in small ways. Sneaking past the guards never came with the feeling that our character was cool or competent. The complications of a partial success always felt like “we screwed up and now the situation is worse.” It all comes down to the framing and narrative interpretation of the dice by the GM. I ultimately think our GM struggles at narrative games.

I'm a big fan of dice pools/bell curves for a number of reasons, but the idea that it's inherently less swingy than linear distributions is absolutely a fallacy. I didn't even realize anybody was claiming that.

Artamo said:

That’s interesting because when I played Scum and Villany (also a FitD game), I had the same feeling but it felt more like everytime we wanted to do something, we failed in small ways. Sneaking past the guards never came with the feeling that our character was cool or competent. The complications of a partial success always felt like “we screwed up and now the situation is worse.” It all comes down to the framing and narrative interpretation of the dice by the GM. I ultimately think our GM struggles at narrative games.

Click to expand...

With partial success games, the group really needs to play a while to get a feel for how often you succeed fully vs succeed partially (which will vary by game depending on normal dice pool size and whether you can do things like spend stress to add dice, etc.). The GM then has to adjust what a partial success looks like in response.

The best example I have is Wildsea -- the difficulty mechanic is such that a challenging task removes a dice (the highest) after you roll. This sounds like a subtle change compared to a more common 'remove a dice before rolling' mechanic, but it is really a lot more punishing, especially if only a 6-success counts as success for relevant purposes. We (my group) played a little while and came to the conclusion that 'the developer just has to be assuming that a partial success is the standard metric of success if attempting tasks under any kind of adverse conditions is at all expected (which it seems to be).'

Artamo said:

That’s interesting because when I played Scum and Villany (also a FitD game), I had the same feeling but it felt more like everytime we wanted to do something, we failed in small ways. Sneaking past the guards never came with the feeling that our character was cool or competent. The complications of a partial success always felt like “we screwed up and now the situation is worse.” It all comes down to the framing and narrative interpretation of the dice by the GM. I ultimately think our GM struggles at narrative games.

Click to expand...

This is the issue with “success with complications” games. It’s a fine balancing act for the GM to come up with complications that are meaningful but do not overwhelm the success.