Dice Probability Distributions: The 3d6 Bell Curve vs the Flat 1d20 Line
Why rolling 3d6 yields predictable, consistent outcomes while 1d20 produces pure volatility, and how the Central Limit Theorem shapes tabletop game design.
1. Discrete Uniform Distribution: The Flat Line of 1d20
In tabletop roleplaying games such as Dungeons & Dragons and Pathfinder, the iconic twenty-sided die (1d20) governs nearly every attack roll, saving throw, and ability check. From a mathematical perspective, rolling a single fair twenty-sided die is governed by a Discrete Uniform Distribution.
Every integer between 1 and 20 has an identical probability of occurrence: P(X = k) = 1/20 = 5.0%. When plotted on a probability graph where the x-axis represents the outcome (1 through 20) and the y-axis represents the percentage chance, the result is a perfectly flat, horizontal line.
Because every outcome is equally likely, the variance of 1d20 is exceptionally high (Var = 33.25, standard deviation approx 5.77). An extreme catastrophe (rolling a natural 1) is just as likely to occur as a mediocre roll (rolling a 10) or a heroic critical triumph (rolling a natural 20). Together, extreme boundary results (1 or 20) consume a massive 10.0% of all rolls.
On a 1d20, a master swordsman has the exact same 5% probability of fumbling their weapon (natural 1) as an untrained peasant. The flat line treats triumphs and disasters with equal indifference.
2. The Bell Curve and Discrete Convolution: The Sum of 3d6
Contrast the flat line of 1d20 with the sum of three six-sided dice (3d6), the core resolution mechanic of systems like GURPS and Hero System. While both mechanics generate results within comparable ranges (1-20 vs 3-18), their probability curves could not be more radically opposed.
When you roll three independent dice and sum their values, the resulting distribution is determined by Discrete Convolution. There are 6 x 6 x 6 = 216 equally likely microscopic permutations, but they map onto only 16 possible sums (3 through 18).
There is only 1 permutation that produces a sum of 3 (1+1+1) and only 1 that yields 18 (6+6+6), giving each an occurrence probability of just 1/216 approx 0.463%. Conversely, sums in the center can be formed in dozens of combinations: 10 can be rolled in 27 distinct ways (12.5%), and 11 can also be rolled in 27 ways (12.5%). Combined, the median numbers 10 and 11 account for 25.0% of all rolls.
In 3d6, rolling the maximum (18) happens once every 216 rolls (0.46%). On a 1d20, rolling a 20 happens once every 20 rolls (5.0%) — making critical hits nearly 11 times more frequent on a d20 than an 18 on 3d6.
3. The Central Limit Theorem: Why Stacking Dice Forms a Bell Curve
The transformation from a flat distribution to a peaked curve is not an accident: it is a direct demonstration of the Central Limit Theorem (CLT). The theorem states that when independent random variables are added together, their normalized sum tends toward a Gaussian (normal) distribution, regardless of the underlying distribution of the individual variables.
Rolling 1d6 is a uniform flat line. Rolling 2d6 introduces a triangular distribution, peaking sharply at 7 (6/36 = 16.67%). Adding a third die (3d6) smooths the sharp peak into a graceful bell curve with standard deviation approx 2.96. Adding more dice (such as 4d6 or 10d6) causes the discrete points to hug a continuous Gaussian bell curve with near perfection.
As more dice are pooled, the probability mass clusters ever more tightly around the mathematical expected value (mean). For 3d6, the mean is exactly 10.5. More than 67.5% of all rolls land strictly between 8 and 13. Wild outliers become exponentially suppressed.
Summing dice replaces linear randomness with clustered predictability. Every additional die added to a pool pulls the distribution closer to a true bell curve and crushes the odds of extreme outliers.
4. Game Design Implications: Chaotic Drama vs Consistent Competence
The choice between a single die (1d20) and a multi-die pool (3d6) is one of the most fundamental design decisions in tabletop mechanics, dictating the narrative tone of the game:
1. **The Swing of 1d20 (High Drama & Swingy Chaos)**: A flat distribution means that fixed modifiers (like +1 or +2) have a constant linear benefit (each +1 always increases success chance by exactly +5%). However, because variance is large, luck frequently overshadows skill. Highly experienced heroes will still fail simple tasks 5% of the time, creating slapstick blunders and unexpected cinematic comebacks.
2. **The Consistency of 3d6 (Simulated Realism & Competence)**: On a bell curve, modifiers have non-linear, dynamic value. Improving an attribute modifier from 10 to 11 grants a massive +12.5% increase in success probability, whereas moving from 16 to 17 grants only +1.4%. Characters who are skilled at a task will virtually never fail under normal conditions, creating a gritty, realistic, and predictable world where strategy and training dominate raw chance.
// Discrete Convolution Algorithm for Dice Probability
function calculateDiceDistribution(diceCount: number, sides: number) {
// Base case: 1 way to get sum 0 with 0 dice
let distribution = new Map<number, number>([[0, 1]]);
for (let d = 0; d < diceCount; d++) {
const nextDist = new Map<number, number>();
for (const [currentSum, ways] of distribution.entries()) {
for (let face = 1; face <= sides; face++) {
const sum = currentSum + face;
nextDist.set(sum, (nextDist.get(sum) || 0) + ways);
}
}
distribution = nextDist;
}
const totalOutcomes = Math.pow(sides, diceCount);
const probabilities = new Map<number, { count: number; percentage: number }>();
for (const [sum, count] of distribution.entries()) {
probabilities.set(sum, {
count,
percentage: Number(((count / totalOutcomes) * 100).toFixed(2))
});
}
return probabilities;
}Frequently Asked Questions
Why do modifiers (+1, +2) feel so much stronger in 3d6 systems than in 1d20?
In 1d20, a +1 modifier always equals a flat +5% chance of success anywhere on the spectrum. In 3d6, because outcomes cluster near the center, a +1 bonus near the median (such as shifting your target from 10 to 11) increases your success chance by a massive 12.5%, making small bonuses feel dramatically more impactful.
Is 3d6 strictly better than 1d20 for game design?
Neither is universally superior; they serve different genres. 1d20 is ideal for heroic, pulpy adventures where surprising critical hits and dramatic failures keep tension high. 3d6 is preferred for tactical, gritty, and simulationist games where players want reliable mastery and fewer fluke failures.
How does Ranomio ensure unbiased dice rolls?
Ranomio generates every die roll using the Web Crypto API (crypto.getRandomValues) combined with mathematical rejection sampling. This completely eliminates modulo bias and PRNG predictability, guaranteeing mathematically pure, unmanipulated rolls whether rolling 1d20 or 3d6.