Dice Roller

JavaScript's built-in `Math.random` is a pseudo-random generator tuned for speed and statistical spread in things like animations or shuffling a UI list, not for resisting an attacker who wants to predict or reconstruct its output, and in some engines its internal state can be partially inferred from a handful of observed values. `crypto.getRandomValues`, the API this tool uses instead, draws from the operating system's cryptographic random source — the same one used to generate encryption keys — which is built and audited specifically to resist that kind of prediction. For a casual dice roll the practical difference rarely matters, but for anything used to settle a real stake — a bet, a competitive game, a decision with consequences — a properly unbiased and unpredictable source is the honest baseline rather than an unnecessary upgrade.

Converting a random integer into "one of N faces" by taking its remainder when divided by N introduces a subtle bias unless N evenly divides the range the random number was drawn from — a d20 in particular does not evenly divide the roughly 4.29 billion values a 32-bit random integer can take, so a naive modulo would make some faces come up very slightly more often than others. This tool instead draws a random value and, if that value would fall into the small unevenly-distributed remainder, discards it and draws again, repeating until a value lands in the evenly divisible portion of the range — the same rejection-sampling approach used by the decision wheel and coin flip on this site, so every face of every supported die is exactly, not just approximately, equally likely.

Tabletop games describe a roll in a compact shorthand — "2d6" means roll two six-sided dice and add their results together, "1d20+5" means roll one twenty-sided die and add a flat 5 afterward, and the die name itself (d4, d6, d8, d10, d12, d20, d100) refers to its number of faces. Different dice serve different purposes by design: a d20 is standard for resolving actions with a wide range of possible outcomes in games like Dungeons & Dragons, a d6 is common for damage rolls and simple probability questions, and a d100 (or two d10s read as tens and ones) is used when a game wants a roll expressed directly as a percentage. Rolling several dice of the same kind and reading the sum, which this tool supports by setting the count above one, is how nearly every tabletop damage roll and skill check is actually resolved at the table.

A die is only genuinely fair if its center of mass sits exactly at its geometric center and every face has an identical chance of settling upward, and inexpensive mass-produced dice frequently fall short of that in small but measurable ways — pips drilled into the plastic remove more material from high-numbered faces than low ones unless the manufacturer compensates for it, mold seams and rounded corners are rarely perfectly symmetric, and repeated use wears certain edges down unevenly. This is exactly why competitive tabletop players and casinos sometimes use dice that are laser-etched rather than drilled, precision-tumbled to remove seams, and individually weight-tested — a level of manufacturing scrutiny that is expensive specifically because true fairness in a physical object is harder to achieve than it looks. A cryptographically random digital roll sidesteps the entire manufacturing question, since there is no physical shape whose mass distribution could ever be uneven.