Ultimatum games and shared narrative control

Narrative control is the degree to which fictional authority is shared between referee and non-referee players in a tabletop roleplaying game. This is one of many properties useful for categorizing and understanding games. Traditionally, narrative control is centralized in the person of the referee but can also be shared either informally based on social norms or formally using game systems. For example, in the Fate engine players can spend fate points to establish facts in the fiction of the game world.

Spreading fictional authority over multiple people can lead to greater recombinant fictional potential. However, delegating authority also decreases puzzle complexity, challenge, and potential surprise (see Zak quote below for more on this dynamic). Spending a point abstractly to make a door be unlocked does not require any creativity or lateral thinking.

At one level, moving along the dimension of narrative control in game design caters to different player preferences. Some players are more interested in being challenged and solving problems while others are more interested in formally structured shared storytelling. Given a set of clear preferences, groups can tailor systems and practices. However, authority in games, just as in the broader social world, is continuously negotiated, even when formally addressed by laws or rules. That is, there are more settings or levers available for games regarding narrative control than simply picking a point on the spectrum.

Anecdotally, while I am generally more in the traditionalist camp of centralized referee narrative control, on reflection I have noticed that I often both explicitly and implicitly delegate fictional control to non-referee players. For example, see how ratlings became part of my Vaults of Pahvelorn. However, the way I find myself delegating fictional authority entails implicit veto power. Though all players, referee and otherwise, contribute to fictional game outcomes, the referee acts as steward. The responsibilities of stewardship in my games include balancing present play against future play and attending to the engagement of individual non-referee players. While this does not mean I adjust outcomes based on what I predict will give particular players more pleasure, it does control whether I linger on a particular fictional experience or work out fictional imperatives quickly.

Based on this understanding of stewardship, as referee I might ask a player what kind of farm a character grew up on or whether they might have relatives in the current town. It is not against the spirit of the game for a player to use this opportunity to gain some present problem solving advantage. However, the player has an incentive to restrain themselves. The more ambitious, far-reaching, or obviously self-interested the interpretation is, the more likely the contribution is to fall afoul of the referee’s steward responsibilities and be rejected. To clarify, this is not at all about protecting a static, perfect setting from the grubby hands of players or ensuring that a plot conforms to a desired narrative arc. Instead, the approach attempts to harness shared creativity while not sacrificing exploratory potential or challenge.

This process works like the ultimatum game in game theory. In the ultimatum game, two players divide some resources between themselves. Player A proposes how to split the resources. Player B decides whether to accept or reject the split. If B rejects the split, both get zero. Empirically, people in the player B responder role are more likely to reject inequitable splits even though such rejection entails personal monetary cost. After all, even one penny is greater than nothing. Because of this empirical fact, proposers have an incentive to not be perceived as too greedy, even though no proposal is formally defined by the rules as invalid.

Mapping this structure to gaming, non-referee players take the proposer role while referees take the responder role. Even this description oversimplifies, as in practice non-referee players may iterate proposals following referee rejection. That said, potential negotiation is limited in practice as groups will not tolerate perpetual renegotiation. Further, once new facts settle, offhand details may lead to surprising fictional consequences, potentially both advantageous and disadvantageous to player goals. This adds to the richness of the game as players incorporate the fictional logic of more inputs. Like butterfly wings shifting weather patterns.

Comment from Zak on the effect of narrative control locus on challenge:

the problem for me with a lot of player-created content ( as a GM ) is the fact that what I then give them then has less of a chance of being a surprise and less of the character of a puzzle. If what’s there has even a 25% chance of being what they decided would be there every time then that’s 25% less fear and dread and giddy anticipation.

The problem for me as a player is that I don’t get surprised, it’s less of a puzzle, it’s less challenging for me (the more info I have, the easier a challenge is), it bores me (I can create content whenever I want when I run a game, why should I do it when I’m playing?) and it robs me of the specific challenge of “If I want something to exist in the game I have to find a way to build it”.

When I’m asked “so, Zak’s PC, what’s over the ridge” my immediate response is “all the treasure int he world and the big bad’s head on a spike”–not because I don’t like making stuff up but because when playing I am conscious that I am trying to direct all my mental energy to exploiting every affordance to get specific goals done that could take years . Asking me to then turn to Author stance is just asking me to do a much easier job with much lower stakes that is consequently less fun.

(Click this link to return to pros and cons paragraph above.)


Forber, P., & Smead, R. (2014). The evolution of fairness through spite. Proceedings of the Royal Society of London B: Biological Sciences, 281(1780), 20132439.

11 thoughts on “Ultimatum games and shared narrative control

  1. dwbapst

    I think the gradient of narrative control isn’t a matter of player preference, but differences among roleplaying games. How much narrative control I want as a player depends on whether I’m playing Mountain Witch, or L&L. And, in not all games would it be appropriate for the referee player to veto a player’s narrative addition; rather the social dynamics of the group at large, and wanting to encourage a fun narrative thread will keep in line, rather than an explicit ultimate veto from a particular participant.

    1. Brendan Post author


      Another way of saying that is that you do not have a strong preference regarding degree of narrative control. You are comfortable delegating that dimension to the game designer.

      Other players may have stronger preferences about this dimension and because of that preference may like (or dislike) games that concentrate or spread around narrative control.

      1. dwbapst

        Sure, yeah. I guess the way I see it, what games we choose to play in real life depends on a large number of variables, and I’m not sure level of comfort with narration is often a primary concern. For example, whether you think your friends will like playing this game with the space aliens or not, or how long a typical session will take, or whether you feel at that moment more like dealing with a bloody Viking family drama or solving a dungeon-lord’s preposterous puzzles… At best, you could say ‘whether players have narrative control’ is just as important, and doesn’t necessary override other factors.

        I will admit I have a few friends who dislike narrating anything, and rule out playing any game that involves players narrating anything, ever. They’d rather sit out the game.

        And so, I’d argue game choice is the causal factor, and that comfort with narrative control might not really be separable from the context of a particular game. Player comfort with narrative control might even vary considerably from game to game. For example, they might dislike the explicit-ness of using FATE points, but once you tell them to pretend its all a TV show in Primetime Adventures, then framing scenes and introducing background characters might just click for them and be fun for them.

  2. Michael Prescott

    Nice post, I particularly enjoy the connection you’ve made to the ultimatum game.

    I think there’s another factor in the traditional style, which I see as generally about deliberately avoiding OOC negotiation of game facts.

    On the one hand, the players aren’t supposed to author outcomes in the world (they stick to their character’s emotions, thoughts and attempted actions), but I think GMs have a corresponding avoidance of intimate PC details, which would presume or impinge on the emotional make-up or sense of identity of the PC. (e.g. “A man steps out of the crowd, he’s a dead ringer for your missing twin brother whom your mother asks you about every year.”)

    If the GM were to say that, suddenly the player is thrust into an OCC conversation. They didn’t see their PC as a twin, and nobody had mentioned the mother before. Now, a bunch of facts about the twin brother are relevant. Who makes those up?

    When the GM tacitly steers clear of this gray zone around the PC, they’re participating in keeping the PCs as potential murderhobos, people without strong connections to the people/groups/places they meet. That’s not a judgement (I’m fine with muderhobos), I bring it up because I don’t often hear about the GM’s role in reinforcing “murderhobo-ness.”

    My personal preference is to let players propose facts that are useful to them, but quantify the likelihood of the connection existing and being usable. (With a GM veto, to boot.)

    One of the cool things about it is that it collapses this gray zone by avoiding the ultimatum game. If you want to sneak into the Bishop’s house, it would be neat to know his washer woman and have her sneak you in. Say that happens to be plausible (your character was raised in this city district), but the Bishop is careful about screening all his hires for entanglements (or more likely, ensures he’s got a shitload of leverage on them), so the GM sets the odds at 25%.

    This is especially fun when complications are a possibility for borderline or failed rolls, such as the washer woman knowing you but being resentful, traitorous, etc.

    1. Brendan Post author


      By in-character versus out-of-character, are you thinking of actor versus director stances?

      I think that narrative control is distinct from stance in that stance can be about fictional character knowledge versus player knowledge where narrative control is not necessarily concerned with fictional character knowledge at all.

      For example, a player could have in-character-relevant narrative control (this person is my brother) or out-of-character-relevant narrative control (the door in this house that I have never visited is actually unlocked because I spend a fate point).

      I don’t think proposing useful facts with referee veto entirely collapses the ultimatum game though, it just moves the game to the proposing step. That is, the step that leads to engaging the random resolution procedure. One could also model the resulting odds as a measure of the proposal’s fairness: a less fair proposal results in the referee setting poorer odds. So there is an incentive to make propose smaller asks because that minimizes risk. I also like that approach though and think it works well. Further, using randomness as the final arbiter increases the perceived impartiality of the ruling as the player could attribute failure to the dice rather than the referee’s decision.

  3. rogerginersorolla

    The advancement of narrative stakes and their acceptance or rejection on the grounds of plausibility is a feature of the “matrix game” approach, which codifies conventions around a tradition of freeform gaming that stretches back into the prehistory of wargames.


    It naturally depends on the good faith, shared assumptions, and willingness to both be convinced and doubt of all parties. Any ruling during play has an element of this, where the player proposes something, the “digital” underpinning of the rule is not clear, and the referee and player resort to “analog” argumentation to simulate what happens on the fly.

    The James Wallis game Baron Munchausen is perhaps the earliest “story game” apart from this tradition to use this concept.

    1. Brendan Post author


      I hadn’t heard about that tradition before, so thanks for the link. In current terms, it seems like the matrix game approach removes the referee from the generation of the ruling.

      In both cases, it seems like the social norms around proposing and evaluating are central to system functionality. Good faith and so forth are certainly a part of required norms, but I wonder if we can’t identify other, more concrete elements useful for making systems that do not require comprehensive rules function better.

  4. Paul T.

    This is a really interesting and thoughtful post. I agree with it in most respects. Very nicely put together!

    My one criticism would be that it lacks a little bit of larger-scope thinking. For instance, it becomes clear that it’s not as simple as “distributed authority decreases challenge” once you step outside the bounds of a typical referee and players relationship. What if we design a set of random tables together, with everyone contributing, and then use those in play? Or a team or writers put together an adventure module, which another group of people play? Or we agree to index a set of books, and use those indexes to generate monsters and treasure for our game on the fly? I’ve participated in an excellent old-school D&D campaign where most players had input into the setting and even took turns running adventure sessions.

    As you point out, in practice everyone delegates some narrative authority to the other people at the table; the trick (which pretty much everyone intuitively can understand) is to police how that narrative authority is exercised. You, as the referee, writing a detailed background for my character wouldn’t hurt the game but is extremely untraditional, and likely rejected on grounds of ‘being a strange thing to do’. A player writing in some detail about the cultural mores of goblin tribes could be very welcome, on the other hand.

    At a table where everyone implicitly or explicitly understands that facing fictional challenges is the creative goal of our play, people instinctively react badly to a player using narrative authority to undercut that interest. I can describe (in great detail!) the physical appearance of my horse, its history, its personality, and so forth. However, if in that description I start to give the horse wings or poisoned hooves, the other players will cry foul. It wasn’t a question of narrative authority which made them cry foul; it was a particular invention which threatens to shake the foundation of the way we set up, overcome, and arbitrate challenges in the game. A referee who uses her “narrative authority” to improvise details of the fiction clearly intended to sway in-play outcomes (whether to the detriment or to the benefit of the players!) from what appears to the group to be “rightly earned” is just as suspect, and as likely to be objected to.

    In this light, it’s easy to see that it’s not about referee/GM or players or the ability to “improvise and make things up”; rather, it’s the discipline we must exercise as participants to maintain the legitimacy of the fictional challenges we come to the table to face.

    Given even a slightly different orientation or focus of “what we came here to play”, different styles and types of narrative improvisation can suit the game just as well.

    These things can be wide-ranging, including the rules we use at the table. For instance, if our ruleset, instead of being about combat or exploration, is all about interpersonal relationships and the challenge in play is swaying other characters to your side (imagine, for a second, an RPG with similar play goals to a game like “Mafia/Werewolf”), giving your horse wings in a freeform description wouldn’t cause anyone to bat an eye – that doesn’t make you any more likely to ferret out the spy in our midst, so go for it!

    1. Brendan Post author


      If the challenge is bypassing the trap, then the relevant fictional elements might be gear carried, immediate physical configuration of the environment, and so forth. If the challenge is social maneuvering, then the relevant fictional elements are associations, existing factional alliances, character attitudes, and so forth. So I think your final point about interpersonal relationships fits into the basic structure of the original post.

      I might summarize the topic in your second paragraph as: shared, potentially higher-order challenges that the play group together attempts. I need to think more on this, but I have two immediate reactions. A) There is a difference between in-fiction, diegetic challenges—such as find the treasure, kill the dragon—and meta challenges—such as create the most engaging fictional metropolis or create the best set of random tables together. B) The first kind of challenge seems more like playing a game while the second kind of challenge almost seems more like designing a game. Reflecting, I agree that I may be somewhat artificially constraining the definition of challenge.

      1. Paul T.


        I may have explained myself poorly – my apologies.

        I have no issue with your use of terminology (and I think the sense of your use of “challenge” is very clear here, given the context).

        What I’m trying to point out is that “shared narrative control” isn’t the most significant factor here. It’s the pursuit of diegetic challenge, rather, which creates this dynamic.

        No one is going to worry about – or object – to a player exercising “narrative control” over elements of the fiction which are irrelevant to the challenges being confronted. It’s only when that player starts to move into relevant fictional details that people will sense that a line has been crossed. In other words, describing my character’s clothes, in any detail (and maybe even how people react to them!), is absolutely no problem, and no one will blink an eye. However, if I start to describe that clothing as including plates of armour, or a jet pack, or a ring a magical protection, play will stop and I will be called out. I have moved from describing fictional detail to setting up elements which will affect how we play out upcoming challenges in play. (You’ll note it’s much worse if we’re, indeed, on the brink of or in the middle of a challenge where this would matter! It might be just fine to say that your character wears a red hat, most of the time, but if you’re infiltrating a society where wearing a red hat is a symbol of power, it’s no longer a detail you can just invent freely.)

        It’s even easier to see from the other side: let’s say that the adventure involves entering a dungeon environment where survival hinges on bringing exactly the right quantities of gear and provisions. If the PCs forget to bring a waterskin, they will die of thirst; if they don’t have any garlic, the vampires will kill them – stuff like that. The GM/referee knows the module well (or designed it) and is curious to see how well they will do.

        The PCs step into the dungeon and the magical gate slams shut behind them. They’ve got three mules laden with pack bags, with all their provisions, but we realize that there is a problem: we forgot to write down what was in them (perhaps we never established it at all, or the player who wrote it down isn’t here, or something like that).

        The players turn to the GM/referee and say, “Hey, since I don’t remember, I’ll just start rifling through the bags. What’s in them?”

        This isn’t in any way a breach of “narrative authority” – generally, it’s the referee’s job to answer such questions. (“I don’t know what’s in the bag. I open it. What’s there?”)

        However, the GM/referee now has some really awkward and/or difficult choices to make. Will she include water in the provisions? Garlic? Doing so will largely decide what the PCs manage to accomplish in the dungeon, and whether they can get out alive.

        Her deciding the contents isn’t a breach of narrative authority – not at all – but it is a breach of the game contract, so to speak. The referee is now “playing for both teams”, nullifying any real handling of diegetic challenge in at least some portion of the upcoming adventure.

        This is why, in such cases (as with the player who claims his character is wearing a red hat) most groups who play in this style will introduce a method or procedure for handling this kind of thing – so the players and/or referee don’t have to decide arbitrarily. Perhaps we will pause play, assign a monetary amount, and have the players “buy” all their provisions all over again, retroactively. Perhaps we will randomize it. Perhaps we will use ability checks to see if the characters were “smart enough” to bring certain things – anything to lend an air of objectivity and stop people from “putting their hand on the wheel” and threatening to nullify the nature of upcoming challenges.

        I hope that helps clarify my point.

      2. Paul T.

        Perhaps a more concise version of my point here might be:

        This is a game design issue, not a “narrative authority” issue. The narrative authority is an important piece of the puzzle, but it’s not the only thing at play here.

        If facing challenges is a large part of the desired gameplay, then the way you distribute narrative authority at the table must support that. This can be done in fairly distributed ways (as it was in that multi-GM hexcrawl campaign I mentioned) or in an extremely top-down way (I have a friend who ran a game like this and the players didn’t even have any authority over their characters’ backstories, for instance).

        The key is that the moment-to-moment procedures of play all support the creative goals of the game; once that’s on the table, lots of different formats are possible.

        I like that view, and find it more useful or broadly applicable, than a more traditional dissection of “player proposes, GM disposes”. It gives us more flexibility and better tools as designers, GMs, referees, and players. It gives us a more robust understanding of what’s actually happening at the table.

        This was an excellent mini-essay, and very interesting to read. I’m just pointing out the bigger picture! Thanks for humouring me.

Leave a Reply