There are dozens of different ways dice have been used in RPGs, and we are likely to see many more in the future. This is not an evolution from bad methods to better methods - there is no such thing as a perfect dice-roll system suitable for all games (though there are methods that are suitable for none). But how will a designer be able to decide which of the existing dice-roll method is best suited for his game, or when to invent his own?
There is no recipe for doing this - it is in many ways an art. But like any art, there is an element of craft involved. This column series will attempt to provide some tools and observations that, hopefully, will give the reader some tools for the craftsmanship involved in the art of choosing or designing dice-roll mechanisms for RPGs.
Overview
Ever since Dungeons & Dragons was published in 1974, randomization has (with a few exceptions) been a part of role-playing games. Randomization has been used for creating characters, determining if actions are successful, determining the amount of damage dealt by a weapon, determining encounters (``wandering monsters'', etc.) and so on. We will look mainly at randomizers for action resolution - the act of determining how successful an attempted action is. The reason for this is that action resolution is in many ways the most critical part of an RPG and the part that is hardest to get right.
Dice of various types are the most common randomizers in role-playing games, and D&D was indeed known for introducing non-cubic dice into modern games. A few games (such as Castle Falkenstein and the Saga system) use cards as randomizers, and some "diceless" games, like Amber, use no randomizers at all, apart from the inherent unpredictability of human behaviour. We will focus on dice in this column series, but briefly touch on other randomizers.
This first installment will head out by discussing some aspects of action resolution that it might be helpful to analyse when choosing a dice-roll mechanism.
Aspects of action resolution
When a character attempts to perform a certain action during a game, there are several factors that can affect the outcome. We will classify these as ability, difficulty, circumstance and unpredictability.
- Ability
- is a measure of how good the character is at performing the type of action he or she attempts. This can be a matter of natural talent, training and tools. The quality of the ability will typically be given by one or more numbers, such as attribute, skill, level, weapon/tool bonus, feats or whatnot. This can be modified by temporary disabilities or favours such as injury, fatigue, magic or fate.
- Difficulty
- is a measure of how hard the action is (independently of the ability of the person attempting it). This can be in the form of active opposition, inherent ``hardness'' or a combination thereof. This is usually also given as one or more numbers.
- Circumstance
- is a measure of external factors that may affect the outcome, making it harder, easier or less predictable. This can be terrain, time of day, lunar phase, weather and so on. Often these factors are modeled as modifiers to ability or difficulty, but they can also be modeled separately.
- Unpredictability
- Unpredictability of an outcome may be either due to inherently random factors or due to factors that are non-random but unknowable or uncontrollable. The degree of unpredictability often depends on the type of action performed - if the character tries to beat an opponent in a game of Poker, the outcome is more random than if the game was Chess, especially if there is a large difference in ability between the two opponents. But the outcome may be unpredictable even in situations with no explicit random element, such as when throwing a ball through a hoop.
We will now look at some properties that action resolution systems might have. In my opinion, a game designer should think about these properties, even if only to conclude that they are unimportant for his or her particular game.
Detail and complexity: So, is the best action resolution mechanism the one that models the above aspects most realistically or in most detail? Not necessarily. First of all, more realism will usually also mean higher complexity, which makes your game more difficult to learn and play, and more detail will typically mean more categories (of skills, tasks, etc.) and larger numbers (to more finely distinguish between degrees of ability, success, etc.), which will require larger character sheets and more calculation. Nor is utmost simplicity necessarily the best way to go - the result may be too inflexible and simplistic for proper use.
So what is the best compromise between simplicity and realism/detail? There is no single answer to that, it depends on the type of game you want to make. For a game that is designed to emulate the silliness of Tex Avery cartoons, the fifty-fifty rule (regardless of ability, difficulty and circumstance, there is 50% chance that you will succeed in what you try) is fine, but for a game about WW2 paratroopers, you might want somewhat finer detail. Nor does detail and realism have to be constant in a single game - if the game wants to recreate the mood in The Three Musketeers, it had better have detailed rules for duels and seduction, but academic knowledge can be treated simplisticly, if at all. On the other hand, if the game is about finding lost treasure in ruins of ancient civilizations, detailed representation of historic and linguistic knowledge can be relevant, but seduction ability need not even be explicitly represented.
In short, you should not decide on the details of the action resolution mechanism before you have decided what the game is about and which mood you want to impart.
Interaction of ability and difficulty: In some games, ability and difficulty (including aspects of circumstance and predictability) are combined into a single number that is then randomized. In other games, ability and difficulty are separately randomized and the results are then compared, and you can even have cases where ability and difficulty affects how the randomization is done. Similar issues are whether preactive or reactive actions (e.g., attack versus defense) are treated similarly or differently, whether opposed and unopposed actions are distinguished, and how multiple simultaneous or chained actions are handled.
Degrees of success and failure: In the simplest case, all a resolution system needs to determine is "did I succeed?", i.e., "yes" or "no". Other systems operate with degrees of success or failure. Degrees can be numerical indications of the quality of the result, or they might be verbal characterisations such as "fumble", "failure", "success" and "critical success". Systems with numerical indications usually use the value of the dice roll more or less directly as degree of success/failure, while systems with verbal characterisations can use a less direct mechanism to identify extreme results.
Nonhuman scales: Some games, in particular superhero or SF games, operate with characters or character-like entities at scales far removed from humans in terms of skill, size or power. Such games need a resolution mechanism that can work at vastly different scales and, preferably, also handle interactions across limited differences in scale (large differences in scale will usually make interactions impossible or trivially one-sided).
Luck versus skill: Let us say we set a master up against a novice. Should the novice have any chance at all, however remote, of beating the master? In other words, shall an unskilled character have a small chance of succeeding at an extremely difficult task, and shall a master have a small chance of failing at a routine task?
The luck versus skill ratio may depend on the task in question -- some tasks are inherently more random than others, such as Poker versus Chess.
On a related note, the amount of random variability in a task may depend on ability. In the "real world", you would expect highly skilled persons to be more consistent (and, hence, less random) than unskilled dabblers, but, as we shall see, this is not true in all systems.
Hiding things from the players: A GM might not always want to reveal the level of ability of an opponent to the players until they have seen him in action enough times to make their own judgment. Similarly, the difficulty of a task may not be evident to a player before it has been attempted a few times, and the GM may not even want to inform the players of whether they are successful or not at the task they attempt (if, e.g., they want to figure out if someone is lying).
Regardless of which system is used, the GM can roll all dice himself and tell the players only as much as he wants them to know. But players often like to roll for their own characters, so you might want a system where the GM can keep, e.g., the difficulty level secret, so the players are unsure if they succeed or fail or by how much they do so, even if they can see the numbers on their own dice rolls.
Diminishing returns: Many games make it harder to improve one's ability the higher it already is. This is most often done through increasing cost in experience points of increasing skill or level, but it may also be done through dice: A player "pays" (either by exercising the skill or by spending a fixed amount of experience points) for a chance to increase a skill. Dice are rolled, and if the roll is higher than the current ability, it increases. Such mechanisms are used both in Avalon Hill's RuneQuest and in Columbia Games' HârnMaster.
Alternatively, you can have linear cost of increasing ability, but reduce the effectiveness of higher skills through the way abilities are used in the randomization process, i.e, by letting a dice-roll mechanism give ever decreasing benefits for added ability.
The shape of things to come
The next installment will give an overview of basic probability theory as it applies to dice-rolling. After that, we will look at and analyse various classes of dice-roll methods. Then we will look at some themes, such as scale, that cut across these classes. We will also look at the history of dice and dice as physical objects, in particular which shapes of dice you can have and still expect them to be fair. We will also look at some alternatives to using dice.
