One of the great innovations of roleplaying game design came in 1977 when Steve Perrin and his Runequest design friends realized the advantage of using the same mechanic for attacks and skill checks. One core mechanic replaced the jumble of different mechanics that D&D used to determine if an attack landed, if a lock opened, if a secret door was spotted, and so on. Core mechanics brought a simplicity and consistency that makes RPGs easier to learn and play. Now even D&D follows Runequest’s example. Attacks work almost exactly like ability checks and saving throws.
But do roleplaying games play better when attack rolls work differently from other ability or skill checks? After all, D&D characters tend to make far more attack rolls than ability checks. And unlike most attack rolls, single ability checks open doors—literally and figuratively, and steer the course of an adventure.
Some newer games break the pattern of making checks and attacks the same. In Draw Steel attack rolls and ability checks differ, mainly because checks can fail, but attacks always succeed enough to do damage. The attack roll determines the amount of damage dealt. Even in D&D, attack rolls and ability checks work a bit differently. A natural-20 attack always hits and scores a critical, but the king disregards your natural-20 Persuasion roll. He still won’t give you his crown.
Treating attack rolls and ability tests differently recognizes that the two types of checks differ in key ways.
Frequent rolls even out highs and lows
A d20 roll yields extreme numbers like 1 and 20 as frequently as middle numbers, creating unpredictable results and adding excitement. D&D swingy d20 attack rolls play well because players make far more attack rolls than other tests. The high number of attack rolls tends to even out the random swings of each roll. Characters sometimes miss, but they hit often and still contribute to the fight.
For ability checks, that d20 roll creates a different dynamic where the roll of the die weighs much more than the +5 attribute bonus of the most exceptional characters. In D&D worlds, the mighty barbarian fails to open a pickle jar, and then hands it to the pencil-necked wizard who easily opens the lid. Such outcomes feel wrong. (See Why D&D’s d20 Tests Make Experts Look Inept and How to Make the Best of It.)
Depending on your distaste for a game like D&D that adds surprises at the price of making experts look inept by routinely letting them fail easy checks, the game might play better with something like a 3d6 roll for resolving ability checks. D&D has a precedent for checks based on a bell-curve die roll. Before D&D added ability check mechanics, some gamers used this house rule: You succeed if a 3d6 roll totals less than your relevant ability score. So a PC with an ability score 11 succeeded half the time, while someone with a 17 almost always succeeded. You can quibble about aspects of this mechanic, but it made high and low scores decisive enough for these traits to show in play. A similar mechanic would never work for to-hit rolls. Characters would virtually always hit, making attack rolls perfunctory. In 2004, the third-edition Unearthed Arcana book proposed a rule variant substituting a 3d6 roll for the d20. “The bell curve variant rewards bonuses relatively more and the die roll relatively less.” A D&D-like system might find an ideal compromise by using d20 attacks and either 3d6 or 2d10 ability checks. Such a compromise would lose the elegance of a single core mechanic, but original D&D used different mechanics for everything and the game thrived, so perhaps we could adapt to two.
Rolls where nothing happens
Repeating the same attack can be fun. Even a missed attack only brings momentary disappointment; the game speeds on to another turn. In contrast, an ability check that fails may create an inconclusive result that invites another try, perhaps starting a tiresome series of attempts that go nowhere and stall the game. Many newer games try to break such patterns. “In Daggerheart, every time you roll the dice, the scene changes in some way. There is no such thing as a roll where ‘nothing happens,’ because the fiction should constantly be evolving based on the successes and failures of the characters.
“A ‘failure’ should never mean that a character simply doesn’t get what they want, especially when that would result in a moment of inaction. Every action the players take should yield an active outcome—something that changes the situation they’re in.”
D&D assumes that ability checks work using the same principles as attacks, so as with an attack, a character can keep trying the same check. For both attacks and checks, circumstances might force a character to stop trying, but otherwise players keep rolling until they get the result they want. One check just represents one attempt of what could be many, and the outcome of a roll might be that nothing happens. Roll again and hope for a 20. The process can become so tiresome that third edition invented the “take 20” rule to skip to any mathematical inevitability.
Checks that focus on intent
D&D-style games tend to treat checks and attack rolls as the indifferent physics of the game world. Imagine an omnipresent DM rolling a DC 1 dexterity check every time a creature crosses the room. (With a swingy d20 roll, that means someone like me with a -1 Dexterity modifier fails to cross 1-in-20 rooms.) Intentions, drama, and fun hardly matter. Creatures can climb half their speed, making checks every turn, until they reach the top or fall to their death.
Games like Burning Wheel break the pattern retries by making checks start with a player declaring what they intend to accomplish, and then letting a roll decisively answer whether the character succeeds at their intent. “A player shall roll once for an applicable test and shall not roll again unless conditions legitimately and drastically change.” To reach a decisive answer, the difficulty of the check depends on intent. So a test to see if a character can pick a lock becomes harder if the character wants to work quietly, and harder still if they want to finish quickly before the guards return.
This attention to intent lets a failed check result in more natural consequences than a temporary failure and “I want to take another Utilize action to try again.” Instead, something behind the door hears the tampering and opens the door. Or the guards come into view the moment the lock pops. Either way, instead of stalling at a locked door, the game races ahead. Even if the check only reveals that the character fails to open the lock, the result is decisive and the players learn the lock surpasses their skills. Instead of trying again, they must find another route. Maybe just ring the bell and bluff. The story continues.
Such decisive checks assume that the character makes their best effort to succeed under the circumstances. You can’t try picking the lock with your non-dominant hand first, so that you gain a second attempt using your dominant hand. You can’t get a second attempt by oiling the lock. Your rogue almost certainly knows more about picking locks than you; they already tried the oil.
Focusing on intent also enables one check to apply for an entire task. Burning Wheel calls this the “Let it Ride” rule. Especially with that swingy d20, a series of checks nearly guarantees a failure. The expert climber falls and the master sneak breaks wind. Letting it ride means that one roll decides whether a character reaches their intended goal unless conditions drastically change.
You might protest that a good DM can run checks this way in D&D. Exactly. That’s the brilliance of focusing on intent and letting checks give decisive results. Even though the D&D rules steer toward a different style of play, DMs can opt for a decisive approach that works better.
Degrees of success and failure
Except for critical hits and a few powers with effects that get worse “if the saving throw fails by 5 or more,” D&D tests never show degrees of success or failure. D&D started without crits and only includes them now because players love them. Few players realize that crits hurt their characters more than monsters. Monsters typically die in three rounds. If they take a crit, that makes two. PCs may take hundreds of hits and numerous crits over their career. One crit could end a long run. Fumbles used to be as common a D&D house rule as crits, but when gamers realized that the skilled warriors with multiple attacks also rolled the most natural ones, fumbles fell from favor. D&D co-creator Gary Gygax knew crits and fumbles hurt PCs more than monsters and fought to keep them out of the game.
Other games have check mechanics that do show success and failure. For example, Fate includes a range of 11 adjectives from “terrible” to “legendary” describing degrees. Many gamers love the storytelling potential of a range of potential outcomes. The challenge of such mechanics comes when game masters keep facing the task of inventing the benefits of a “superb” success and the consequences of a “poor” failure. Also, as a player who enjoys the tactical decisions of combat, I favor games where I know the potential results of my attack rolls and rarely wonder what the game master might invent.
Most RPGs with a big combat pillar tend to tabulate the possible outcomes of good and bad attack rolls. A strong attack can always deal extra damage. In the heyday of critical hits and fumbles, games included grisly tables of potential outcomes and characters took to tying weapons to their hands to avoid all the dropped swords. None of this benefited the fighters who made most of the attacks or the PCs in general, who suffered from every monsters’ limb-severing crit.
If ability checks with degrees of success and failure tax the creativity of game masters, and if crits and fumbles punish PCs more than their foes, then how can degree of success and failure work?
Daggerheart stands out for letting bad attack rolls hurt attackers, but only with mild and indirect consequences: Initiative passes to the players’ foes and the game master gains Fear points to boost monsters. The Hope and Fear points that come from good and bad combat rolls just power characters and their foes.
For most play styles, fumbles on to-hit rolls don’t work. Pathfinder 2 includes critical successes and failures for both ability checks and saves, but the game skips fumbles for attack rolls. This stands as an instance where different rules apply to different types of checks. Modern games where players can fumble attacks either tend not to feature combat or tend to have characters that make equal numbers of attack rolls.
Pathfinder tries to ease the GM’s creative load by specifying the results of critical success and failure for most saves and actions. For instance, a critical success on a save might mean that you take no damage instead of half. And a critical failure on an attempt to pick a lock means you break your tools. Nonetheless, without factoring intent into the consequences, Pathfinder has limited possibilities for outcomes. How many times will the rogue break a lockpick before they just start packing 20 extras?
Decisive checks that factor intent work better because they lead to more natural consequences and because they never stall the game with inconclusive results and a pile of broken lock picks.
Different mechanics
Draw Steel shows a state-of-the art example of attack rolls and ability checks that use different mechanics. Ability checks show decisive answers with degrees of success. Bad failures bring complications that raise tension. Attack rolls skip fumble results, but include degrees of success that replace separate damage rolls. This streamlines the attacks in a D&D-style game, and cutting perhaps 20 seconds from each of a hundred attack rolls made during a session leads to a faster pace. As for the advantages of a 3d6 bell curve versus a d20 swing, Draw Steel splits the difference with a 2d10 triangle.
