D&D General What is player agency to you?


If time permits, can you point to a post that's part of that discussion?
I first mentioned the Whitehall study here: https://www.enworld.org/threads/what-is-player-agency-to-you.698831/page-137#post-9097886

I mentioned it again here: https://www.enworld.org/threads/what-is-player-agency-to-you.698831/post-9103462

@Campbell referred to sociological notions of agency here: https://www.enworld.org/threads/what-is-player-agency-to-you.698831/post-9107784

I don't agree that there is one single definition within sociology. I've seen multiple different definitions within management theory - which is a discipline of sociology.
Sure, there are debates in sociology over what the precise phenomenon is that should be studies, is worth trying to give rise to in organisational structures, etc; and also debates about the best way to characterise that phenomenon.

I still think there is merit in contrasting the sociological use with the use in philosophy of action. Eg if my interest, from the perspective of management theory, is how to cultivate agency in at least some employees so as to better achieve the organisation's mission (a practical example I have in mind, though it predates modern management theory, is the training of German compared to English junior officers in the First World War), then I am not going to get very much out of going to a philosophy seminar about Anscombe or Davidson on "basic actions".

Is it right then that what you are focused upon are actions players take without reasons for taking them?
No. I'm not focused on actions the players take in general at all. I mean, we could look at a table of RPGers and note (eg) that their breathing is autonomic, that their drinking and eating of snacks is largely reflex/habitual, that one of them who is on a diet is deliberately refraining from eating snacks, that one participant deliberately raises their voice for dramatic effect in delivering a line, etc.

In this respect there is probably no interesting difference between a group of RPGers, a group of boardgamers, or even perhaps a group sitting around completing a jigsaw together. But when I talk about player agency in RPGing, as I've made clear from post 211, I mean the agency that the player exercises over the content of the shared fiction. And in my view it is a truism that this agency can vary in degree, depending on the game rules and associated procedures of play. And in no way does it contradict this observation to observe that giving a player the full freedom to just tell a story would mean that there is no longer a game being played.

Agency is the objective and subjective capacity to exercise self-determination through action and effort. In the context of an RPG, agency is the objective and subjective capacity to perform actions and make decision that have a meaningful impact on the outcome of the game. Note that the perception of what constitutes outcome and meaning differ between individuals.

I think it is important to not reduce agency to simply concern itself with the decision space of the player in the context of the game. One has to consider the effect of the decision. In order for agency to exist, the player must not only have a meaningful and open-ended freedom of choice in what actions to perform and which decisions to make. to act. The actions and decisions must also be able to have the desired impact.
I think I've been pretty clear about this throughout the thread. I have focused on what I regard as the distinctive outcome of RPG play: collective creation of a shared fiction. And I have talked about agency in respect of the capacity of a player to affect that.

In this thread we've mostly discussed narrativist systems and how they contribute to agency. And they certainly do for a lot of players. The extent to which they provide agency is of course directly linked to how important the player considers the sharing of narrative authority and the decision space for impacting the shared fiction. Some players do not have a desired to impact the shared fiction beyond controlling a character. Earlier in this thread a poster asked me what I meant by "Story Mechanics" - and to me that refers to mechanics that allow the player to impact the shared fiction by means other than direct control of a character. In classic D&D systems I would consider character creation a Story Mechanic by this definition, and the same goes for character progression mechanics like levelling up. So Story Mechanics are not a matter of whether they are "gamey" or not.

But if we return to how agency can relate to other concepts than shared narrative authority and Story Mechanics
What I have quoted here seems to rest on a frequent but wrong presumption namely, that mechanics that allow the player to impact the fiction only by means of controlling their PC are not suitable for delivering a high degree of player agency in respect of the shared fiction. The reason that that presumption is wrong can be set out by pointing to the RPGs that refute it: Apocalypse World, Burning Wheel, Prince Valiant, just to name a few. AW has almost no "story mechanics" (and the ones it has are optional elements of particular PC builds), Prince Valiant has none, and Burning Wheel can easily be played without the player having to do anything but describe what their PC does, thinks and feels. (As I've given examples of upthread.)

Or to put it the other way around, it is possible to have a RPG that follows the D&D paradigm in respect of what the players have control of, when it comes to impacting the shared fiction and yet to give the players a very high degree of agency in respect of the shared fiction. As I have posted repeatedly upthread, the way that this works is by imposing obligations on the GM when it comes to narrating other elements of the fiction beyond what it is that the PCs do.

Freedom without agency is a very common pattern within sociology, and it repeats within RPGs. For example, if all roads lead to Rome, and I don't want to go to Rome, then it doesn't matter how many roads there are. But on the other hand, if what matters to me is the journey, then the choice of road is highly relevant.
The most typical example of this in RPGing, I think - or at least the paradigmatic example - is illusionism of the sort typical in CoC play, and in much D&D play especially if informed by the post-DL and 2ned ed AD&D ethos. This has also been discussed in the currently active "random tables and player agency" thread.

In this sort of play, the players are free (within limits of good taste and social harmony) to introduce colour and characterisation around their PCs. As a group, they may choose for their PCs to take the high road or the low road. But the GM will manipulate background fiction, perhaps also dice rolls, to ensure that the pre-planned events more-or-less come to pass. Given the sort of agency that I am interested in (as explained just above), I regard this as rather low player agency RPGing.

a fair number of players that I've met over the years are mostly concerned with "fun/entertainment" and/or character progression. The former is sometimes derisively consider a casual and amateurish priority - where players don't care about the integrity of the shared fiction, don't care about their ability to impact it, etc. and are thus little more than passive consumers of low-brow entertainment. It is not entirely untrue for a subset of these players - for some people RPGs are just a light-hearted pastime. They can still have plenty of agency - if they're able to meaningfully act and interact to make the game more entertaining. For example, such a player might feel deprived of agency when playing with a group that is too serious or too concerned with following the rules of the game. To an observer they may seem devoid of agency when playing with like-minded individuals, but that's not the case - because there is a decision space and a meaningful ability to impact the outcome of the game.
I feel that, in this paragraph, exercise of agency is being equated with having a good time. Also, I think, that *non-game
play social agency* (such as cracking jokes that (i) don't actually advance the play of the game, and hence (ii) irritate serious participants) is being blurred with the play of the game.

I accept that the lines here aren't perfectly clear, but we wouldn't argue that snakes and ladders becomes high agency as a game because a group of people enjoy sitting around rolling the dice, laughing at one another's wins and losses, cracking jokes based on the pictures on the board, etc.

For the character progression crowd - they're much more focused on whether progression delivers tangible increases in mechanics-based character power and/or narrative character identity, than on the shared fiction or the interactions at the table. Again, this kind of play is sometimes regarded as primitive and boring by those who value other aspects of RPGs. But it is a valid kind of agency. A power-gamer may feel deprived of agency if character progression does not allow for synergistic use of mechanics - or at the very least goes beyond a simple scaling of character power. Such a player does not care about whether they can affect the narrative - but they do care about whether they can pump out bigger numbers or feel the thrill of executing some cool combo of mechanics to dominate combat.
For me, this raises a different sort of question, along the lines of "what is the game"? Is it the creation of the shared fiction, or is it the more individual act of manipulating the PC build mechanics? I've got my own strong preference in this regard.

Mateas and Stern go on to reiterate that

..the effect must relate to the player intention
This seems fairly obvious. But it does have fairly devastating implications for many of the ideas of agency one sees put forward in relation to RPGing, namely those which hold that it is enough to show significant player agency that a player's declared action for their PC will produce an outcome that matters to the player, while accepting that all the intentions will be those of the GM.

In RPGing, assuming a fairly traditional allocation of participant roles, for the effect to relate to the player intention requires the GM to be under exactly the sorts of constraints I have been mentioning since post 211.

They develop their theory via a concept of material and formal causes. Material causes are resources present in a game such as mechanics. Formal causes are goals or ideals toward which things are directed. Players introduce a source of formal causes, distinct to games. They come to conclude that

A player will experience agency when there is a balance between the material and formal constraints... An imbalance results in a decrease in agency.

As such, they characterise agency as inhabiting a "sweet spot" in design space. Sufficiently afforded and constrained on both layers.

a player in an interactive drama becomes a kind of author, and thus, as an efficient causee, contributes both materially to the plot and formally to elements at the level of character on down. But these contributions are constrained by the material and formal causes (viewed as affordances) provided by the author of the interactive drama.
For those reading along who aren't familiar with the ideas, the notions of material, formal and efficient cause go back to Aristotle:

* Material cause refers to the stuff that something is made of (eg the statute is made of clay);

* Formal cause refers to the arrangement or organisation of the thing (eg the statute is of a woman) - when the statue is reshaped, its form changes although its matter does not;

* Efficient cause refers to what we most often think of causation in our own mechanistic/scientistic framework (eg the cause of the statute is the sculptor).​

Aristotle also has a fourth notion of cause - teleological cause - which is the reason that something happens. This is controversial in Aristotle's usage, because he imputes these sorts of reasons to non-thinking things (eg the teleological cause of the acorn is the oak tree it will grow into) - these authors, as summarised, appear to bleed the notions of formal and teleological cause, and also perhaps the notions of material and efficient cause. (Caveat: I haven't read the paper, and am relying on the summary.)

If I was going to apply the Aristotelean schema to the creation of a shared fiction via RPGing, I would consider "story elements" as material causes - who gets to decide on these? (Consider eg the role of relationships in Burning Wheel or Torchbearer.) I would consider thematic connections as formal causes - the manner in which the material causes are organised (eg who is friend and who is foe?). The efficient causes would be the processes of play whereby the participants are empowered to introduce new material causes and/or shape the forms. Teleological cause would be the "agenda" as Ron Edwards et al thought of it - ie why are we all gathered together doing this thing, rather than (say) going for a walk together, or playing bridge?

log in or register to remove this ad


Fundamentally, the game is all ritual. It's a set of ritualized movements that make my family spend a couple hours around a table together, provides a topic of conversation but does not require conversation about itself, and slides easily into the background when the real goal is to spend time together. It is a fantastically low agency game, the outcome of which is almost entirely determined by chance, barely inching out something like War or Go Fish.

It's possible to discuss agency in that context because the game has a known, shared goal and you can measure player impact on achieving it with the available actions, but the meta-goal of the game, the ritual purpose if you will, has little to do with the goal of play, and is if anything, enhanced by having a needlessly complicated system that provides very little agency.
I've often posted, over many years, that of the whist family of games I prefer five hundred to bridge, and of the classic board games I prefer backgammon to chess, because both my preferred games involve more luck and also demand less of the player (in terms of cognition and play skill) to do well.

To put it simply, they are more light-hearted.

This doesn't mean that they are high agency, however, either in general or relative to some contextual pemertonian matrix of agency. I play them because they're fun and relaxing, not because they allow me to exercise my agency is a game player. My interest in exercising agency as a card or board game play is actually quite finite. My preferred venue for high agency game playing is RPGing.


Given that all games limit agency to some degree in order to be a game - then there should be games with more agency than PbtA or burning wheel or etc (and if they don’t currently exist then they theoretically do so as we can all imagine them).
This claim is not self-evident, and is in fact doubtful.

A game has rules - to use the Suits notion that @clearstream favours, it involves the conventional/consensual/voluntary, self-imposition of obstacles to achieving the goal of the game. This is done because there are payoffs from the experience, other than simply achieving that goal. For instance, RPGing with a traditional allocation of participant roles produces a shared fiction with characters, including protagonists, and potentially with rising action, *climax and resolution.

Obviously the most efficient way to generate a shared fiction of that sort is to sit down and write it. Doing it via a game, like a RPG, places self-imposed obstacles in the way because the experience that results is fun. One reason it's fun is because, as a player, all I have to do is declare actions for my PC. (Note that I reiterate here my view that "story mechanics* are largely irrelevant to the core of the discussion about player agency in RPGing.) Another reason it's is because, as a GM, all I have to do is imagine compelling situations for the protagonists, without having to write out how those situations resolve. The resolution is punted to the system.

Is there a way of preserving the game of the RPG, including via the traditional allocation of participant roles, while increasing the degree of agency enjoyed by players over the shared fiction beyond what is achieved by Apocalypse World, Burning Wheel, and some other RPGs that emerge from the same set of design ideas and intellectual milieu? I'm not persuaded that there is: those systems, in different technical fashions, fully deploy the device for bridging player agency across participants roles so as to achieve player agency. That device, as I first posted in post 211, is that players

establish their own goals and aspirations for their PCs (including working with the group collectively to establish the appropriate backstory and setting elements to underpin those goals and aspirations), and then the GM relies on those goals and aspirations as cues for their own narration of framing and consequence​

What device, consistent with the RPG allocation of roles, are you suggesting would allow even greater player agency over the content of the shared fiction?

In short the posters saying they prefer higher agency games aren’t actually looking for the highest agency possible (shouldn’t be controversial). Instead they are content with the amount and types PbtA and others offer them. It’s not simply a desire for high agency - it’s also a desire for particular types of game rules.
No. It's a desire to play a RPG as opposed to, say, a conch-passing game. Eero Tuovinen wrote about this over a decade ago: The pitfalls of narrative technique in rpg play


I feel that there are reasonable grounds to say that an expert referee would do something that the dice cannot, such as include the effects of things in the fiction that the system as written is insensitive to. And I have on my mind additional merits such as where a referee is communicative, consistent, judicious. Very often in other domains I will prefer a person's decision over a random one.
This prompted two thoughts:

(1) If there is stuff in the fiction that you want to matter, and your current mechanical framework isn't making it matter, then you need a better framework! (I think this is a version of @Manbearcat's point upthread.)

(2) In many areas of human life, it is understood that there is a difference between allocating opportunities via (say) turn-and-turn-about, or via (say) the toss of a coin, than via the decision of another person. Doubly so when the other person is also one of those to whom opportunities are being allocated!

The point of the dice roll in AW or in BW - despite their important technical differences - is fairly similar: it establishes who gets to take the lead in having a say over the shared fiction. This also often involves establishing constraints that govern whoever gets the next say, so the agency of that person who got to take the lead ramifies into the circumstances where they no longer have the lead.

Replacing the randomiser with the decision of a spectator would be a non-negligible change. (Consider how controversial this would be as, say, a change to the rule for determining which captain gets to make the bat/field call in a cricket match.)

Replacing the randomiser with the decision of a game participant is a change that it seems to me quite obviously and profoundly alters the distribution of agency.


(He, Him)
A game has rules - to use the Suits notion that @clearstream favours, it involves the conventional/consensual/voluntary, self-imposition of obstacles to achieving the goal of the game. This is done because there are payoffs from the experience, other than simply achieving that goal. For instance, RPGing with a traditional allocation of participant roles produces a shared fiction with characters, including protagonists, and potentially with rising action, *climax and resolution.
Something to consider is that what counts as "fiction" is potentially diverse. Some play aims to produce dramatic fiction (as I think you are acknowledging with "potentially".) Much or most play aims to produce protagonist fiction. Other play can aim to produce the fiction of what it's like to be the character (and nothing beyond that!) Lyric games aim to produce a poetic fiction. These are mixable and scalable, and there may be further alternatives.

Some commentators divide "story" from "narrative", in order to imply that the former is tighter and more constructed - deriving more from linear forms of narrative - while the latter is looser... meanings that a person can read-off or construct for themselves. Several have started bridging the narratological/ludological divide by speaking in terms of micro-narratives, for example pointing out that "ludo-narrative dissonance" cannot be anything but "narrative-narrative dissonance", for if there is no implied narrative in the ludo part, from whence the dissonance? Postclassical narratology lets in highly diverse possibilities for fictions or narratives.

It could be most accurate to cut it at - TTRPGing produces a shared fiction with characters.
Last edited:


(He, Him)
I feel like those last seven words illustrate the nexus of your (what seem to me to be Sim-Immersionist) priors and where a lot of disagreement stems from.
I failed to communicate my intended postulation, as you have read it in an unexpected way. I had in mind a comparison between system in videogames and system in TTRPG. In both, player moves must be interpreted into system as (mainly) triggers and settings (of parameters). They can also be interpreted to assemble lists of results, if those are not fixed. I didn't have in mind any comparison between two specific game texts, nor simulationism or immersionism.

In videogames, the system has a large but generally finite range of moves it is sensitive to. There are a couple of ways that one could view the link between fiction and system in videogames. One way is to observe players self-narrating what they see and do, and translating that internal fiction into a secondary "language" supplied by the game controls. Another might be to suppose that said language itself is sufficient to sustain meaningful fiction (here I'm thinking of the "micro-narratives" postulated by postclassical narratologists and ludologists.) Either way, it could be said that players supervene meaning upon the system-language.

So the concept of "sensitivity" I'm thinking of is a technical one: it's what the system "looks for" or "can detect". The "messages" a system subscribes to and has means to interpret. It should be obvous, but TTRPG systems aren't really sensitive to anything: a person considers fits between fiction and system along whatever lines they think the game text and meta intends.

As to "what they think the system intends", one can observe degrees of certainty which means the probability that two different persons will enact it in the same way. I) It is very probable (yet still not certain) they'll enact system to system inputs the same way. II) There will typically be a small set of explicitly exemplified "snippets" of fiction that when presented exactly, will be enacted the same way (for a PbtA move, the exact examples of fiction that the move "looks for".) III) There will be a vast and infinitely varied set of possible fiction that lead to uncertainties in enactment.

So by "sensitive" I mean whatever is covered under I) and II). I don't offer either agreement nor disagreement about that: my response to @EzekielRaiden (my edited version, that is!) was intended to be limited to proposing a thought-experiment. Anything not prefixed with wording such as "my position is", is likely something I have no fixed commitment upon. I shall take more care with language in future, to avoid misunderstanding.

Good systems are written with the intention of being sensitive to some things and insensitive to others!
100%... albeit I would take that to sustain rather than dispatch the proposal above.
Last edited:


(He, Him)
I know I already replied, but this required a more focused response rather than my general one. Inconsistency would actually be much worse than random. Dramatically so. My "meddling Professor" analogy is also a thing where an unknown inconsistency is worse than pure randomness.

If something is random, you can learn how to prepare for randomness. Especially the kind of randomness usually used in TTRPGs, since there are tools available to the player to influence the probabilities. There will still be times where chance simply lies against you. That's part of life, and part of gaming. Statistics is literally the science of accounting for randomness so that we can still learn from it.

If something is genuinely both inconsistent and secretive, you can't learn to prepare for it. There is no pattern, by the definition of inconsistency, but players cannot know how or why the inconsistency occurs. Certainly it is possible for the inconsistency to be partially occluded by chance, e.g. the black box decrees something just secretly fails, and it just so happens that the roll would have failed anyway. That sometimes chance will coincide with the inconsistent input does not mean the inconsistency is irrelevant.

Indeed, I would argue the reverse of what you did here: inconsistency would surely be no better than random--and likely much worse. Try to "learn" from inconsistency, and you are essentially guaranteed to develop false beliefs, because (in the relevant sense) there are no correct beliefs about it. That follows directly from the fact that it is inconsistent.
I'm really not seeing how you can tell inconsistent from random, when it comes to the string of values that will index results. The intuition you have seems to be that on observing some series of numbers, given true random I have whatever odds of predicting the next value, whereas with inconsistency whatever I predict will be defied somehow.

These "false beliefs" will turn out to be true beliefs about the sorts of number sequences come out of the black-box. Thus, they represent learning about its behaviour. There is a hidden term in your thought-experiment is that the student isn't really trying to understand the black-box, they're trying to understand a secondary phenomena that you have postulated. To fit with my thought experiement, the black-box must contain both that secondary phenomena and the professor, and learning must be targeted at that gestalt.


(He, Him)
But you still have the players and (in most systems, in particular) the GM using the system, which does change the "baseline" of the system. For example, a GM can deprive players of agency in any kind of system by making a certain type of scenario. This can be done even purely using prepared materials and no direct railroading. You can devise sandboxes that regardless of how impartially you adjudicate the game, the players have low agency when acting within them. And the opposite holds too - a system which is comparatively low agency can be turned into a session of massive agency if the GM hosts a session that gives players tons of agency by virtue of the scenario and the interactions.
I would add principles to @FrogReaver's potted description. The situation you are describing to the extent it has denied players of agency is due to mismatching principles. The deprivation is felt in influence over outcomes, where process is also an outcome. But this is as to within-game "major premise" agency. So you could instead be referring to agency oriented toward given outcomes, i.e. within-game "minor-premise" agency.


Follower of the Way
I'm really not seeing how you can tell inconsistent from random, when it comes to the string of values that will index results. The intuition you have seems to be that on observing some series of numbers, given true random I have whatever odds of predicting the next value, whereas with inconsistency whatever I predict will be defied somehow.

These "false beliefs" will turn out to be true beliefs about the sorts of number sequences come out of the black-box. Thus, they represent learning about its behaviour. There is a hidden term in your thought-experiment is that the student isn't really trying to understand the black-box, they're trying to understand a secondary phenomena that you have postulated. To fit with my thought experiement, the black-box must contain both that secondary phenomena and the professor, and learning must be targeted at that gestalt.
But the "gestalt" as you put it isn't real--and the player/student is deceived into believing that the black box is solely the real deal. That there are no, as you put it, "R-C dice," but just dice.

It is not possible to have true beliefs about something inherently inconsistent (at least, with regard to the parts that are inconsistent). An inherently inconsistent thing contains contradictions. You cannot believe a contradiction (I don't ascribe to paraconsistent logic, and even if I did, such logics are actually more restrictive than ordinary logic, not less!), but an inconsistent structure necessarily contradicts itself. Sometimes, a belief about it will be true, and other times, it will be false. No pattern can arise which clearly favors one or the other, lest it become consistent.

In other words, you cannot learn about an inherently inconsistent thing. It will defy learning. As long as the black box is the actual origin of control, but the "R-C dice" are the apparent locus of control, learning is not possible.

The instant you remove the secrecy, meaning the black box is clearly known to be the actual locus of control, then the situation is resolved; players know the inconsistency is present and can attempt to address it (though, TBH, there's not much you can do, given its black-box nature.) And, likewise, if the black box is removed, then even if secrecy remains in other parts of the process, the situation is again resolved. Players will be correctly reasoning along the lines "I took action X, and Y resulted, therefore action X caused Y to happen," and can thus avoid X if Y was undesirable, or pursue X if Y was desirable. Public randomness (the dice, plainly visible) simply means that we must add the caveat "I successfully took action X," but since the success is purely public knowledge, there is no issue with this modification.

Remove ads


Remove ads

Upcoming Releases