Design properties
Each mechanic is coded on eight closed-vocabulary properties describing its canonical form in isolation, not any particular game that contains it. Coding guidelines are below each property; the full rubric is on themethodology page.
Player agencyordinal
How much the outcome of one invocation varies with the player’s choice at that moment.
| Value | Meaning | Mechanics | Share |
|---|---|---|---|
| low | Outcome mostly fixed by prior state or chance. | 60 | |
| medium | One meaningful choice among a small, constrained set. | 92 | |
| high | Open choice space where selection dominates the outcome. | 39 |
Coding guideline
Low: outcome mostly fixed by prior state or chance (loot tables, saving throws). Medium: one meaningful choice among a small constrained set (push your luck: stop or continue). High: an open choice space where selection dominates outcome (worker placement, drafting).
Randomness
Whether chance is resolved before the decision (input), after it (output), both, or not at all.
| Value | Meaning | Mechanics | Share |
|---|---|---|---|
| none | No randomiser in the canonical form. | 125 | |
| input | Chance resolved before the player decides. | 34 | |
| output | Chance resolved after the player decides. | 29 | |
| both | Both stages are intrinsic to the canonical form. | 3 |
Coding guideline
Input = randomness resolved before the player decides (a dealt hand, dice rolled then allocated, a generated map). Output = randomness resolved after the decision (an attack roll, a loot drop). Code "both" only when the canonical form has both stages.
Interaction
How the mechanic couples players to one another.
| Value | Meaning | Mechanics | Share |
|---|---|---|---|
| solo | Operates identically regardless of other players. | 78 | |
| indirect | Players affect each other only through shared state, without targeting. | 47 | |
| direct | Targeted effects on a specific opponent. | 33 | |
| cooperative | A shared goal is intrinsic to the mechanic. | 6 | |
| mixed | Canonical implementations routinely span two or more modes. | 27 |
Coding guideline
Solo: operates identically regardless of other players. Indirect: players affect each other only through shared state without targeting (blocking a worker space, depleting a market). Direct: targeted effects on a specific opponent. Cooperative: a shared goal is intrinsic to the mechanic (hidden roles is not cooperative, whatever its information coding; team-based is mixed). Mixed: canonical implementations routinely span two or more. Opponents controlled by the game system are not players: a mechanic whose canonical form pits one player against the system codes solo; code direct only when the canonical form targets another human player.
Complexityordinal
Rules weight needed to teach the mechanic on its own, not its strategic depth.
| Value | Meaning | Mechanics | Share |
|---|---|---|---|
| low | One or two sentences, no exceptions. | 84 | |
| medium | A paragraph with a few sub-rules or timing exceptions, or a single lookup table. | 92 | |
| high | Several interacting tables, prerequisite graphs, or interacting sub-rules. | 15 |
Coding guideline
Low: one or two sentences, no exceptions. Medium: a paragraph with one to three sub-rules or timing exceptions; a single lookup table on its own is medium. High: several interacting tables, prerequisite graphs, or multiple interacting sub-rules.
Pacing effect
Effect on the rate at which play approaches an end condition or climax, or on turn length.
| Value | Meaning | Mechanics | Share |
|---|---|---|---|
| accelerates | Snowballs or forces endings. | 50 | |
| decelerates | Lengthens turns or stalls progress. | 23 | |
| neutral | No systematic effect on pace. | 108 | |
| variable | Depends heavily on parameters. | 10 |
Coding guideline
Accelerates: snowballs or forces endings (clocks, end-game triggers a player can pull forward). Decelerates: lengthens turns or stalls progress (large-hand management, hit-point attrition). Variable: depends heavily on parameters (procedural generation, difficulty settings). Neutral: no systematic effect.
Information
Visibility of the mechanic’s own state.
| Value | Meaning | Mechanics | Share |
|---|---|---|---|
| perfect | All relevant state is public. | 131 | |
| hidden | Concealed from everyone. | 18 | |
| asymmetric | Some players know what others do not. | 38 | |
| mixed | Both hidden and asymmetric components are intrinsic. | 4 |
Coding guideline
Scope is the mechanic’s own state. Perfect: all relevant state is public. Hidden: concealed from everyone (a shuffled draw order). Asymmetric: some players know what others do not (fog of war, a private hand). Mixed: both hidden and asymmetric components are intrinsic. State known only to the game system and to no player is hidden; state known to a human referee or game master but not to the other players is asymmetric.
Feedback loop
Whether success makes further success more likely, less likely, neither, or is self-damped.
| Value | Meaning | Mechanics | Share |
|---|---|---|---|
| positive | Success compounds. | 35 | |
| negative | Success is penalised or trailing players are compensated. | 13 | |
| none | No systematic coupling between past and future success. | 134 | |
| mixed | A positive loop with a built-in damper, or both directions present. | 9 |
Coding guideline
Positive: success compounds (engine building, experience levels). Negative: success is penalised or trailing players are compensated (catch-up bonuses, handicap stones given to the weaker player). None: no systematic coupling. Mixed: the canonical form contains a built-in damper (deck dilution in deck-building).
Decision densityordinal
Decision points one player faces in one turn’s use of the mechanic, counted without regard to how much each matters (that is agency).
| Value | Meaning | Mechanics | Share |
|---|---|---|---|
| low | Zero or one decision point per turn. | 98 | |
| medium | Two or three decision points per turn. | 72 | |
| high | Four or more decision points per turn. | 21 |
Coding guideline
Count within one turn of the player using the mechanic. Where the mechanic spans several players’ turns without resetting (an auction from lot to sale, a trick from lead to take), count that player’s decisions until it resets. Where play has no turns, count within one encounter in real-time play, or one move in a puzzle solved alone. A decision point is a moment where the rules give the player two or more legal options. Count decision points, not options: choosing among twenty spaces is one decision point. Count only choices the mechanic itself creates: choices made through other mechanics the game also has do not count, and a mechanic that is a constraint, a tally, or a frame with no choices of its own has zero. A referee’s or game master’s rulings on outcomes are adjudication, not decision points; where the mechanic is a menu of choices for the game master (GM moves), count those choices. Inputs that test execution (timing a jump, steadying an aim) are not decision points; count only choices between routes, targets, or tactics. Where the count varies, code the typical count in the canonical form. Code density without looking at agency; the two often disagree. Low: zero or one (victory points, initiative order, placing one stone in Hex, playing one card to a trick, the clue-giver’s turn in Codenames). Medium: two or three (set collection: which card to take and whether to cash in a set; cover systems: whether to move and to which cover). High: four or more (a deck-building turn of playing and buying cards, spending an action-point budget).