The dice math a digital roller gets right (and a physical set hides)

Why 1d20 is flat and 2d6 isn't

Roll a single d20 and every face from 1 to 20 is exactly as likely as every other face — a flat, uniform distribution. That's part of why d20 systems feel swingy: a 5% chance of a 1 and a 5% chance of a 20 sit right next to each other, and nothing in the math favors the middle. Roll two six-sided dice and add them, though, and the shape changes completely. There's only one way to roll a 2 (1+1) and only one way to roll a 12 (6+6), but there are six different ways to roll a 7 (1+6, 2+5, 3+4, 4+3, 5+2, 6+1). The middle of the range is dramatically more likely than the extremes, so 2d6 behaves like a bell curve while 1d20 behaves like a coin flip stretched across twenty faces.

This is the single most useful thing a dice roller can teach you if you actually watch the individual rolls instead of just the total. Set it to two d6 and roll it thirty or forty times in a row — you'll see 6, 7, and 8 come up constantly and 2 or 12 show up maybe once or twice, if at all. Do the same with a single d20 and every number feels equally plausible, because it is. Once you've seen that pattern with your own rolls, "add more dice of the same size" and "roll one bigger die" stop looking like interchangeable ways to get a bigger number — they produce genuinely different probability shapes, even when their averages land in the same place.

A flat modifier shifts everything; it doesn't reshape anything

Here's the nuance that trips people up: adding a +2 modifier to a 1d20 roll and adding a +2 modifier to a 2d6 roll do very different things to your outcomes, even though both "add 2." On the d20, +2 just slides the flat line two spots to the right — every result is still equally likely, you've just raised the floor and the ceiling by the same amount. Nothing about the shape of your luck has changed, only its location. On 2d6, though, the distribution was already curved before you added anything, so a +2 modifier shifts a bell curve that was already clustered around 7. The practical effect is that a flat bonus on a curved roll pushes your most likely outcomes up by exactly the bonus amount, but it does nothing to flatten or steepen the curve itself — the middle stays the most likely zone, just centered two higher than before.

Where this actually matters at the table is when you're deciding between "give this character a bigger modifier" and "have them roll an additional die." A bigger modifier is a guaranteed, unconditional boost — it helps every single roll by the same fixed amount, good or bad luck notwithstanding. An extra die, by contrast, mostly helps your average and typical rolls (because that's where the added die's probability mass concentrates) while barely moving your worst-case or best-case extremes. If you want a house rule that reliably raises the floor — nobody ever whiffs completely — a flat modifier is the cleaner tool. If you want a rule that rewards consistency and shrinks the wild swings around the middle, adding a die and keeping the modifier the same does more of that work. They are not interchangeable, and a few dozen rolls on a dice roller set to log every individual result will show you the difference far faster than working it out on paper.

Why this matters for D&D-style ability checks

Tabletop RPGs lean on this exact contrast constantly, usually without spelling it out. A d20 ability check with a modifier is deliberately swingy by design — the designers want a level-1 fighter to occasionally roll a natural 20 and land a hit nobody expected, and they want a legendary hero to occasionally roll a 1 and faceplant in front of the whole party. That's the flat distribution doing its job: drama through genuine unpredictability. Damage rolls, on the other hand, are usually multiple dice summed together (2d6 for a greatsword, 3d6 for a fireball at higher levels) precisely because a game doesn't want damage output to be wildly random every single hit — it wants damage clustered near an expected value so combat pacing stays roughly predictable, with the extremes reserved for genuinely exceptional rolls. If you're building or reskinning a house system, that's the lever to pull: use single larger dice where you want swinginess, and sums of smaller dice where you want consistency.

For quick reference on how that plays out at the table, our D&D Encounter Generator pairs naturally with a dice roller when you need to resolve initiative, saves, or improvised checks on the fly without breaking from the encounter you just rolled up.

When a screen beats a physical set

None of this is an argument against real dice — there's a tactile satisfaction to a handful of polyhedrals that a browser can't replicate, and plenty of tables will always prefer them. But a digital roller earns its place in a few specific situations. Remote sessions are the obvious one: when half the party is on a video call, a shared or individually-run online roll keeps everyone honest without anyone holding a die up to a webcam. A lost or incomplete set is another — you don't need a d12 gathering dust in a drawer if you can pull one up instantly, custom side count included, when a homebrew rule calls for something odd like a d7 or a d30. And for teaching, a digital roller that logs every individual roll is genuinely better than physical dice, because you can fire off fifty rolls in the time it takes to physically gather and re-roll a handful of cubes, turning an abstract probability lesson into a visible pile of data.

If what you actually need is a single random number rather than a sum of dice faces — say, picking a number 1 to 100 for a raffle, or settling something with a coin instead of a die — the Random Number Generator and the Random Coin Flip tool both cover that simpler case without the dice-specific mechanics.

Try it yourself

The Random Dice Roller supports d4 through d100, a custom side count for anything in between, up to twenty dice rolled at once, and an optional plus-or-minus modifier applied to the total. Set it to two or three dice of the same size, roll it a few dozen times, and watch the totals cluster around the middle instead of spreading evenly — then switch to a single larger die and watch that clustering disappear. It's the fastest way to feel the difference between adding dice and adding a number, and it runs entirely in your browser with nothing to install.