What the dice are actually doing
Dice pools: many small chances instead of one big one
Counting successes across a handful changes the texture of failure as much as its frequency.
Counting successes across a handful changes the texture of failure as much as its frequency.
What a pool actually does to probability
A single die resolves a question by landing on one number. A dice pool resolves it by landing on several — then counting how many of those numbers clear a threshold. The shift looks mechanical but it runs deeper than that: replacing one roll with many small independent chances changes not only the probability of success but the shape of how that success arrives.
When Shadowrun first deployed large pools of six-sided dice against a target number in the late 1980s, the arithmetic was unfamiliar enough that players wanted printed tables. What the dice were actually doing was producing a binomial distribution: the more dice in the pool, the more the outcome clusters around the expected number of hits. Roll two dice and every outcome from zero to two is plausible; roll ten dice and results near the tails become vanishingly rare. The binomial distribution that underlies this is the same mathematics that governs coin-flipping, and its key feature is that variance shrinks relative to the mean as pool size grows. A character with many dice becomes genuinely more reliable, not just more likely to succeed.
That reliability is the first design payoff. In a single-die system, a highly skilled character and a novice differ mainly in their target number or their bonus — the d20 still lands anywhere with equal frequency. In a pool system, the skilled character's larger pool narrows the band of possible outcomes. Expertise becomes legible in the texture of the roll, not only in its binary result.
Degrees and the texture of failure
The second payoff — the one that changes play more visibly — is that pools yield natural degrees of outcome. Counting successes means a roll rarely returns a clean yes or no. Three successes on a combat roll land a better hit than one; two successes on a social roll might mean the target is swayed but suspicious. Apocalypse World's move system achieves something similar through numerical bands on 2d6 rather than pool counting, but the underlying ambition is the same: replacing the binary pass/fail with a spectrum.
This texture reshapes failure in particular. On a single-die system, failure is a wall — the roll does not clear the target number, play stops or reverses. On a pool, a roll that yields one success where three were needed can still produce something: a partial outcome, a costly win, narrative leverage for the referee. The failure is granular. It can be narrated as "just short" or "barely enough," which is a different fiction than "nothing happens."
Systems built around pool mechanics tend to encode this spectrum directly into their core vocabulary. The World of Darkness family — which shares the binomial pool architecture — distinguishes botches (no successes and at least one critical failure die) from simple failures, making catastrophic outcomes structurally rare but procedurally specified rather than adjudicated case by case. The World of Darkness core design philosophy is visible in how that distinction shapes referee authority: the rules hand the referee a spectrum to interpret rather than a verdict to deliver.
The cost the pool carries
None of this is free. Pools slow the table. Gathering, throwing and reading eight dice takes longer than one roll, and the count introduces cognitive overhead that a single number against a target does not. Some designs solve this with special dice — Genesys and its predecessors use custom faces that collapse the count into symbol-reading — but that solution trades one cost for another: proprietary components and a higher barrier to understanding what the dice are actually doing at the probability level.
The pool also amplifies the design stakes of pool size. If a high-skill character rolls eight dice and a novice rolls two, the gap in reliability is steep enough to raise questions about whether challenge is still meaningful for the expert — a version of the bounded accuracy problem that fifth-edition D&D addressed on a different axis by capping bonuses to a single die rather than expanding the pool. Every choice about what the dice are doing is simultaneously a choice about what skill, risk and dramatic tension mean at the table.
