Who gets to say it happened?
Every roleplaying system is an answer to one question — how a table shares authority over what is true in the fiction.
Every resolution system is a theory of authority
Every roleplaying system is, underneath everything else, an answer to a single question: who at the table is allowed to say that something is true in the fiction? The referee declares the monster is dead. A player announces her character persuades the duke. The dice come up blank and the door stays shut. Each of these is an act of authority — a claim about the shared fiction that the others are expected to accept. The rules exist, in large part, to say whose claims count and under what conditions.
The question sounds philosophical, but its consequences are mechanical and immediate. A game that places authority entirely with the referee is structurally different from one that places it with a roll, or with the player, or with some negotiated overlap of all three. That difference shapes every session before a single die is thrown. It governs what the players are trying to do with their turns, what the referee is trying to do with the fiction, and what the dice are actually for — whether they arbitrate a conflict or merely generate a consequence the referee then interprets.
The historical source of the problem is worth naming. Dave Arneson and Gary Gygax both inherited a design culture from miniature wargaming, where an umpire held authority over edge cases — not because it was philosophically correct, but because wargame rules were never complete enough to handle every situation. The referee inherited a wargame without a manual for what the job was supposed to be. The authority question was never settled; it was improvised, session by session, in Lake Geneva and in every basement that followed.
The three places authority can live
Authority in a tabletop game tends to settle in one of three locations: the referee, the player, or the procedure. In practice, most systems distribute it unevenly across all three, and the interesting design work is in specifying exactly who holds it when.
Referee authority is the oldest form. The referee listens to a player's action, consults whatever internal model she has of the fiction, and declares a result. Rules enter only as constraints — she cannot rule that gravity runs sideways without some prior fictional cause — but within those constraints, her word is final. The appeal is flexibility; the cost is that the fiction is always filtered through one person's mood, preference, and theory of fair play. A reactive referee is not the same mechanism as a permissive one, even if both are running the same rulebook.
Player authority arrived as a deliberate counter-move. If a player's statement about her character's intention is simply accepted as true — if the rules say her character is persuasive because she took a relevant trait, not because the referee agrees — then the authority structure shifts. Apocalypse World, published by D. Vincent Baker and Meguey Baker in 2010, made this structural rather than incidental by routing all significant player actions through moves: when you do the thing, you roll, and the move text tells you what is now true. The referee interprets consequences, but cannot simply veto the result. The move is the authority.
Procedural authority is what happens when neither referee nor player settles the question, but a table of randomness does. The reaction roll is the clean example: rather than the referee deciding how the ogre feels about the party, a two-dice roll produces a disposition the referee then dramatises. The dice are not adding drama — they are removing a decision from one person and placing it in a neutral mechanism. This is a meaningful redistribution of authority even when no one at the table frames it that way. The exploration turn works similarly: by making time procedural — light expires, rations tick down, wandering-monster checks occur at fixed intervals — the game removes the referee's discretion to simply ignore resource attrition when it becomes inconvenient.
What a resolution system is deciding
When a resolution system is designed, it is not only choosing a probability shape. It is deciding who carries the burden of uncertain outcomes and who gets to interpret them. A roll-under system — testing against a stat the player already has — places the calculation in the player's hands and makes the threshold transparent before the roll. A dice pool that counts successes distributes probability differently, and the way failure arrives (close, then steep) carries its own implications for who controls pacing. The design question is not just "how hard is this?" but "who narrates what the numbers mean?"
The tension becomes clearest in edge cases, and edge cases are where authority is actually designed, whether the designers intended it or not. A rule that says "the referee may call for a roll when the outcome is uncertain" is a statement about where authority lives in ambiguous situations — with the referee, who decides when the fiction is uncertain enough to require arbitration. A rule that says "whenever a player acts within a move's trigger, roll" is a different statement entirely: the trigger test belongs to the table, not the referee's judgment.
The OSR tradition reads this tension back into the earliest published games and finds a useful design principle: when the 1974 assumptions are restored — rulings over rules, fictional positioning over skill-check abstraction — the game grants the referee wide authority in exchange for obligations of neutrality. The referee who runs a dungeon as a designed object, populated before play and adjudicated without adjustment for dramatic convenience, is not exercising unconstrained authority. The dungeon constrains her. The procedure constrains her. What looks like referee power is partly the fiction itself holding ground.
Why the question never fully resolves
The authority problem does not dissolve once a game chooses a system; it simply changes shape. A game with player-facing moves still requires someone to decide when the trigger fires. A game with strong referee authority still relies on players accepting the fiction as shared rather than the referee's private story. A procedural layer can settle a question the referee might have fudged, but someone still interprets the result.
What design can do is redistribute authority deliberately rather than accidentally — make explicit choices about which claims require arbitration, who performs that arbitration, and what constraints apply. The difference between a system where the referee's call is final and one where a move result is owed to the player is not a difference in fun: it is a difference in what the probability distribution is actually deciding and for whom. Every rulebook encodes an answer. The question is whether the designers knew they were answering it.
Read next
More in Who gets to say it happened? · the full index