The heart of any tabletop system is its resolution mechanic — the roll that decides whether an action works. Before you write a single skill or stat, the shape of that roll sets the tone of the whole game: how swingy it feels, how much a trained expert can rely on their training, and how much a lucky amateur can steal the scene. Get the dice right and everything downstream is easier to balance.
Flat versus curved: the core choice
Almost every dice mechanic falls into one of two families, and the difference is everything.
| Engine | Distribution | Feel |
|---|---|---|
| Single die (d20, d100) | Flat — every result equally likely | Swingy, dramatic, unpredictable |
| Multiple dice summed (2d6, 3d6) | Bell curve — middling totals common, extremes rare | Steady, competence-driven, few wild upsets |
| Dice pool (count successes) | Roughly bell-shaped, scales with pool size | Granular, tactile, good for “more is better” |
A flat d20 gives every outcome from 1 to 20 an equal 5% chance. That makes results dramatic: a novice can roll a 20 as easily as a master, so luck looms large. A 2d6 bell curve does the opposite — totals cluster around 7, and 2s or 12s are rare (about 3% each), so skilled characters win consistently and upsets feel special. A dice pool, where you roll a handful of dice and count how many beat a threshold, sits in bell-curve territory too but lets you scale difficulty by adding or removing dice; see understanding dice pools for the detail.
Setting target numbers
Your target number (the value a roll must meet or beat) has to match the die you chose. On a flat d20, spacing targets evenly works cleanly: each +1 to the target simply removes one 5% slice, so a DC of 10 is a coin flip, 15 is hard, 20 is heroic. On a curve you cannot space targets linearly, because the probabilities are not evenly spread. Against 2d6, moving a target from 7 to 8 costs far more success chance than moving it from 11 to 12 — the middle is where the mass lives. Plot your success rates first, then set targets to hit the feel you want rather than assuming equal steps.
Why modifiers matter more on a curve
This is the single most important consequence of picking a bell curve, and it is easy to miss. On a flat d20, a +1 bonus is always worth exactly 5% more chance of success, wherever the target sits. On a curve, the value of a +1 depends on where you are: near the centre it can swing the odds enormously, while out at the tails it barely registers.
Take 2d6 against a target of 7 (roughly even odds). A +1 bonus there shifts a large chunk of probability across the line, because you are moving the densest part of the curve. The same +1 against a target of 11 does almost nothing — there is hardly any probability out there to move. The practical upshot: on a bell curve, character bonuses are powerful and should be handed out sparingly; on a flat die they are predictable and can be more generous. A system that gives out d20-sized modifiers on a 2d6 engine will feel wildly unbalanced.
Degrees of success
Modern designs rarely settle for pass or fail. Adding degrees of success — a critical success, a plain success, a partial success with a cost, a failure — gives the same roll far more narrative texture. Curved systems lend themselves to this naturally: because extreme totals are rare, you can reserve the best and worst outcomes for the tails and let the common middle produce the “yes, but” results that drive good stories. On a flat die you can band the range instead — beat the target by 5 or more for a critical, miss by 5 for a fumble — but remember each band is equally likely on a d20, so criticals will feel more frequent than on a curve.
A designer’s checklist
- Pick the feel first. Want heroic swing? Flat die. Want competence to shine? A curve or pool.
- Map your success rates before assigning target numbers — never assume equal steps on a curve.
- Scale modifiers to the engine. Small bonuses on a curve, larger ones on a flat die.
- Decide how failure works. Degrees of success turn a binary roll into a story beat.
- Playtest the extremes — the least-skilled character against a hard task, and the most-skilled against an easy one — to check the maths matches your intent.
The fastest way to sanity-check a new mechanic is to roll it a few hundred times and watch the distribution. Throw your candidate dice in Dice Drop, tally the totals, and you will see the curve you are designing around take shape in front of you.
Try it yourself. Dice Drop rolls real physics dice in your browser — advantage, keep-highest and modifiers built in.
🎲 Roll dice freeKeep reading: understanding dice pools — how counting successes across a handful of dice builds its own kind of curve.