RPG System Families Compared by How They Resolve a Roll

Comparing games by their setting or page count tells you very little about how an evening will go. What decides that is the resolution mechanic: the small procedure repeated a hundred times a session, whose shape determines whether competent characters feel competent and whether failure is embarrassing or interesting.

Six families cover nearly everything in print. The differences between them are not matters of taste dressed up as design. They produce measurably different outcome distributions, and those distributions are what a group is actually choosing between.

The families at a glance

Flat dice and the cost of swing

A flat die treats every outcome as equally likely, which makes the arithmetic trivial and the results volatile. D&D 5e and Pathfinder 2e both resolve this way.

The trained duellist and the cook have the same chance of rolling a two.

Systems built this way lean on modifiers to separate them, and on the referee to decide how often a roll is called for at all, because the more often you roll a flat die, the less expertise shows.

Percentile systems are flat as well, but they present the chance directly on the sheet. A skill of 65 in Call of Cthulhu or RuneQuest succeeds 65% of the time and nobody has to work anything out. That transparency is why the format has survived for decades in skill-led and investigative games, and why players in those games argue much less about difficulty.

Curves, pools and reliability

Summing dice or counting a pool produces a curve, and a curve is a promise that most results will be ordinary. GURPS rolls 3d6 under a target, Traveller adds 2d6, and Shadowrun and World of Darkness count successes in a pool. Half of all 3d6 rolls land in a four-point band. A pool of six dice produces two or three successes more often than not, about 55% of the time. The exceptional result still exists, but it arrives rarely enough to be worth remarking on.

This changes what modifiers mean. On a curve, a single point near the centre can be worth several times what it is worth at the edge, so designers keep bonuses small and players notice every one of them. It also changes pacing: fewer swings means fewer scenes derailed by a single roll, which some groups experience as fairness and others as flatness.

Playbooks and tiered outcomes

Systems that roll 2d6 and split the result into three bands are making a different argument: that the binary of success and failure is the problem. Apocalypse World introduced the shape and Dungeon World carried it into fantasy. On a bare 2d6 the middle band only ties the miss band at 41.7% each, but add the +1 a capable character carries and it becomes the widest at 44.4%, so the most common outcome is a partial success that moves the story on while introducing a cost.

Mechanically this pushes work onto the referee, who has to invent the complication on the spot. In exchange the rules stop needing a difficulty scale at all, because the tiers do that job. Groups who find target numbers tedious often find this liberating, and groups who want to know exactly how hard something was find it maddening.

Ladders, and dice that only nudge

The ladder family goes furthest in the direction of reliability. Fate rolls four dice bearing plus, minus and blank faces together and sums them, which clusters results hard on zero: the most common outcome is that the dice change nothing at all and the rating on the sheet stands. Descriptive rungs replace numbers, so a character is Good or Superb at something rather than a +2 or a +4.

That produces a game where the character sheet is a strong predictor of what happens, and the dice supply variation rather than drama. It suits tables that want competence respected and find swing irritating.

It suits nobody who wants the dice to be the exciting part of the evening, and a group that picks it for the descriptive ratings without noticing the distribution tends to wonder why the rolls stopped mattering.

Choosing without buying six books

Decide two things first. How often should a competent character fail at something they are good at — occasionally, or almost never? And when they fail, should the game stop and reconsider, or complicate and continue? The first answer picks flat dice or a curve. The second picks a target-number system or a tiered one.

Everything after that is setting and page count, which are much easier to judge from a preview than the mechanic is. Reading the resolution chapter of a free quickstart tells you more about a game than the whole rest of the book.

Common questions