D&D General What is player agency to you?

[pemerton]

This is the basic foundation of classic RPGs in general: The game, being a reality simulation, simply can not ever even come close to rules for everything. The solution was simple: have one person be "God". That one person, the GM, gets to decide EVERYTHING in and about the game. It's literately how the game works.

One of the earliest RPGs is Classic Traveller (1977). Here is the rule for the Streetwise skill (Book 1, p 15):

The referee should set the throw required to obtain any item specified by the players (for example, the name of an official willing to issue licenses without hassle = 5+, the location of high quality guns at a low price = 9+). DMs based on streetwise should be allowed at +1 per level. No expertise DM =-5.​

The same page also states the fictional context that underpins this rule:

The individual [with Streetwise skill] is acquainted with the ways of local subcultures (which tend to be the same everywhere in human society), and thus is capable of dealing with strangers without alienating them. (This is not to be considered the same as alien contact, although the referee may so allow.)

Close-knit sub-cultures (such as some portions of the lower classes, and trade groups such as workers, the underworld, etc) generally reject contact with strangers or unknown elements. Streetwise expertise allows contact for the purposes of obtaining information, hiring persons, purchasing contraband or stolen goods, etc.​

So here, from a rulebook published 3 years after D&D was first published, we see a contradiction of your claim about "the basic foundation of RPGs in general".

But how does this other game your talking about work? An RPG where the player decides the impact of what their PC says or does? To me, that sounds like a DM less game......or even more simply: The player can just stay home and write their novel.

I have not talked about any RPG that fits what you describe here.

I've just stated a perfectly functional RPG rule, that supports high agency play. It does not depend on any so-called "alter reality" ability. It does require the GM to have regard to player goals for their PCs in adjudication.

Every example I've seen has the GM still 100% in control of the whole game. Unlike D&D, the game has some vague rules for the GM to follow about 'agency'. But they are vague. So any GM can "follow the rule(s)" and still just do whatever they want. That is what vague rules do.

Even if a player calls the GM out: "I don't like the way you are following rule 23!" how is that any different from in D&D when a player says "I don't like what you did"? In both cases would not a GM just say "you don't have to like it, but I say it happens." ?

I've provided abstract descriptions of various high-agency RPG systems, as well as examples of actual play. Have you read any of them?

 

[pemerton]

Suppose, as a player, your goal in play is to learn what another person is imagining. In that case, you will not want to exercise agency over the content of what they imagine.

An analogue might be a gift: if I would love to receive a gift that reflects someone else's conception of what they would like to gift me, then I don't want to exercise agency over their choice of gift for me. Rather, I want to be surprised and delighted by their choice!

It seems pretty plain to me that a lot of RPGers want to learn what another person is imagining. This is often described using the language of "discovery" or "exploration*. I don't see what benefit flows from describing this in distorting language as the one who is learning also being in control.

 

[FrogReaver]

Though it would be rather an unusual vector, if that were the case. Though perhaps that's my physics perspective talking.

But let us use this vector analogy, really put it to work. The components of the vector are scalar. Either they are boolean variables (1/0, present/absent), or they are not (allowed other number values; amount or degree of something.) In the case of booleans, the analysis is quite simple: more vectors is more vectors, that's all you need.

I've talked through this a few times before, but it's not as simple as you are saying. There's primarily 2 reasons - though I'm just going to focus on just 1 in this post.

Consider just 5 dimensions of agency and 2 players
Player A has agency of <0,0,1,1,0>
Player B has agency of <1.1,0,0,1>

Now suppose the last dimension is actually just a linear combination of the first and 2nd dimensions. In this case player B does not have any more agency than player A despite the most obvious potential metric for total agency being simply counting each dimension that has a 1 for a given Player. In short, counting 1's by dimension is only useful if we've already done our due diligence to ensure the dimensions are all linearly independent. I don't think anyone has came close to showing that. In fact, I'd suggest most of the discussion has already used descriptions analogous to linearly dependent dimensions of agency.

Additionally, consider an infinite dimension vector. One cannot simply count 1's anymore to determine more agency. I'd suggest agency is much more likely to have infinite linearly independent dimensions. In which case, this counting metric also fails.

In short for counting to work as you propose we must ensure we are dealing with finite, linearly independent dimensions.

 

[FrogReaver]

I don't support or particularly want to engage with the weird binary argument, but I think you have missed something. I don't think instances of agency as you're laying them out here are strictly additive, unlike items on a restaurant menu. I was arguing something similar with my attempt to divide out narrative vs. ludic agency early; depending on your stated goal it's possible to add options that increase agency along one axis, while decreasing it as much or more along another.

I think it's interesting to note that if you add options that increase agency along one axis and it decreases options that increase agency along another axis, then we explicitly talking about dimensions that are not linearly independent.

 

[FrogReaver]

For instance, so-called "character agency" doesn't seem to describe a different type of agency at all (it is still referring to agency in respect of the shared fiction); rather, it just describes a particular constraint on the exercise of agency (ie the only direct change in the fiction the player can make is by declaring an action for their PC).

Im curious - Why shouldn't different constraints on the exercise of agency be referred to as different types of agency? Is there something different that 'types of agency' would better refer to?

The fact that Apocalypse World, Dungeon World and Burning Wheel operate primarily via this mode of agency is (it seems) routinely ignored.

I'm not expert but I can think of examples that were provided that unless I'm mistaken were from those games that went beyond what I've called 'character agency'

And the introduction of additional constraints (eg the player can't declare their PC's veridical mental states except by first consulting the GM about permissible contents of those mental states) seems to be typically intended by proponents of "character agency" but it is not normally mentioned by them - eg they will often say the player is in charge of what their PC thinks and feels but in fact that turns out not to be the case, given the GM's control over permissible veridical mental-state contents.

I'm not quite sure what this means - and normally I would attempt to guess but I think that will likely do more damage than just asking for clarification.

How is it like being able to wish something into existence? What "reality" is being altered? The player declares an action for their PC. How does the fact that it is about the PC's past actions change it's character as an action declaration?

I don't have a great way to say this but I'm sure there's a way to -
Playing a scene in the present tense - jumping to the past tense to establish something useful for the present tense (not time travel) - from the perspective of the present tense right before the jump to the past tense there was a change. The mechanical nature of how that change came about matters - mechanics and not just the resulting fiction mattering is a sentiment you've shared with me quite often in the past so it mattering here shouldn't really come as a surprise. It's not really wishing something into existence - but absent ironing out the language to better articulate the issue - it's not like all these diverse people all make up the same issue either.

I'd also suggest this is the same reason that 'remembering the tower with potentially useful stuff in it is nearby' also is an issue for many. Honestly, flashbacks of something not already established in the game and memories of stuff not already established in the game are essentially the same thing. A flashback for all intents and purposes is a memory!

 

[FrogReaver]

Suppose, as a player, your goal in play is to learn what another person is imagining. In that case, you will not want to exercise agency over the content of what they imagine.

An analogue might be a gift: if I would love to receive a gift that reflects someone else's conception of what they would like to gift me, then I don't want to exercise agency over their choice of gift for me. Rather, I want to be surprised and delighted by their choice!

It seems pretty plain to me that a lot of RPGers want to learn what another person is imagining. This is often described using the language of "discovery" or "exploration*. I don't see what benefit flows from describing this in distorting language as the one who is learning also being in control.

You explained the first part eloquently and i liked the post for that.

But taking that analogy and applying it to this style of RPGing as if that's all it is means the analogy substantially breaks down - because players in this style aren't wanting to simply be delighted by someone elses choice - though that certainly is a part of exploration - at least in some sense. More importantly - in exploration, the person exploring is in control of the exploration - even though their exploration does reveal 'delightful gifts' the DM provides.

 

[EzekielRaiden]

I've talked through this a few times before, but it's not as simple as you are saying. There's primarily 2 reasons - though I'm just going to focus on just 1 in this post.

Consider just 5 dimensions of agency and 2 players
Player A has agency of <0,0,1,1,0>
Player B has agency of <1.1,0,0,1>

Now suppose the last dimension is actually just a linear combination of the first and 2nd dimensions.

Then it isn't a dimension in the first place.

That's literally how "dimension," for vector spaces, is defined: the rank of the matrix formed by the basis vectors, which can be determined by looking at them in reduced row echelon form. Any completely null rows mean the rank of the matrix is less than the number of vectors that comprise it: they are not linearly independent. Only the vectors that are linearly independent describe a basis set, and thus, tell you the dimension of the vector space.

In fact, I'd suggest most of the discussion has already used descriptions analogous to linearly dependent dimensions of agency.

Not at all. I consider "character" agency, the way others have described it, to be independent of player agency (though I recognize that Pemerton, for example, would disagree with me on that.) That's why I've called them genuinely different types of agency!

Additionally, consider an infinite dimension vector. One cannot simply count 1's anymore to determine more agency. I'd suggest agency is much more likely to have infinite linearly independent dimensions. In which case, this counting metric also fails.

While it may be the case that a perfect and complete accounting of all possible forms of agency would do that, do you think any game designed by a human being would include not just an infinitude of opportunities (and potentially magnitudes, if we allow non-booleans), but also an infinitude of types of agency?

I don't think humans can design such a thing. Instead, we design for finitely many. I kind of thought that was a given, that humans cannot make a thing that would actually provide rules for infinitely many types on top of instances (and potentially magnitudes.) Tactical infinity is already difficult to design for as is, and that's only on one single axis!

In short for counting to work as you propose we must ensure we are dealing with finite, linearly independent dimensions.

"Linearly independent dimensions" is a tautology in vector spaces. A space's dimension is, by definition, the maximum number of linearly independent vectors you can have, aka, the number of vectors in any basis set for that vector space. As stated, I don't believe a human can make a game that offers infinitely many forms of agency.

 

[Cadence]

Yes it's multi-dimensional. Scalar in the sense that it admits of degrees - ie some people enjoy more agency than others.

Got it. I'm used to scalar valued explicitly meaning a unidimensional real value (at least as far as I know in stat, math, CS, the physical sciences, the definition Google pops up - the only issue is if they want to fight over whether it's the same as being a 1-vwcor or not I guess).

If it can be thought of as a vector with numerical components then some people could enjoy more of one dimension of agency and less of another. And the question of finding a total amount feels like it depends on if the different components are orthogonal/uncorrelated or not (as in orthogonal vs oblique rotations in factor analysis) and if the different components are viewed as being on the same scale in a meaningful sense (is one unit of X agency equal to one unit of Y agency). You could still have A have more agency than B if A was greater on every component.

 

[Cadence]

How is it like being able to wish something into existence? What "reality" is being altered? The player declares an action for their PC. How does the fact that it is about the PC's past actions change it's character as an action declaration?

And this is the discussion I think we had before. I'm not sure what example I used then, but say we have: "An ogre who snuck into the keep steps out from behind the shed and advances on you." "Good, does it fall in the pit I dug their yesterday." It feels like this could be done in good faith in some set-ups, and could also be abused as an I win button by a player viewing it as getting to alter reality if they just use the right phrasing.

Leo in one of the Percy Jackson series has a magic belt pouch (iirc) that he can pull pretty much any mundane tool from. From a game play effect is that different than a character being able to say "oh, I put X in it yesterday" and being able to pull it out whenever they need that X? Did Mary Poppins really pack her carpet bag in advance?

<Like I said, I'm pretty sure we've been down this road before, albeit with different particular examples, so no need to redo it here unless you really have a burning desire too. If we haven't, and I were the site admin with connections at google, I guess I could always go insert them in a thread somewhere and ask for an insta-crawl to hide my tracks. And yes I realize that's different for you because it would be changing your past - but as long as you don't have total perfect recall it should still fly with any of the other readers here who weren't intimately involved at the time.>

 

[Maxperson]

I'm not sure the comparison is as apt as you might like. Especially because the way we actually discuss complexity has a better tool: the number of spots on the board (squares for chess, intersections for go.) Chess has 64 squares. Go has 361 intersections (19²). Further, there are far more valid go boards than there are chess boards, because (for example) it is perfectly valid to have a board with no white stones, but a board with no white chess pieces is not valid. Estimating reasonably, there are about 10^170 valid go boards: each intersection has three valid states--empty, black, white--and there are 361 such intersections, so 3^361 = 1.7^172, whittle off a power of two for good measure. The typical estimate for the possible number of chess games (not just board-states) is "only" about 10^120, and the number of possible boards (including invalid states) is estimated to be only about 10^45. Go is vastly more complex purely because of its board size. Even a novice board is 9²=81 intersections, giving 3^81 or about 10^38 possible board-states. A mathematical estimate of the number of actually valid game-states is around 2*10^170.

Go is more complex to code because you have to account for far, far more states than you do in chess.

So....yeah. You can in fact say that one game is more complex than another because it has more options; you just have to be careful about how you approach that, recognizing that "options" means more than just "distinct pieces." Go has more options by virtue of the size of its board. Chess simply makes it seem like it has more states because it has more pieces, but with far fewer places for them to be, that advantage is quickly irrelevant.

Go has one single action. Put a piece on the board. You can "choose" which spot to put the piece on, but you are railroading into the single action the DM has forced you into. Put a piece on the board so the game moves forward like the DM wants.

Chess on the other hand may have fewer squares, but the choices are real. No railroading. I can pick which piece to move. I can move it more than once or not at all. There is no DM forcing the game and giving an illusion of choice.

DM = Rules in these examples.

Thus, it seems you must deny one of the following:

1. that there can be kinds/forms/types of agency at all
2. that moments/situations/events/instances of agency can be counted at all
3. that "agency" is a concept with any meaning whatsoever
4. that "narrative" games offer equivalent character-agency to "trad" ones (or "neotrad" or whatever, I don't mean to hold you to a label here.)

1 is manifestly false, as players, authors, creators, etc. clearly have the ability to influence or control things within their creative space that ordinary humans do not within their existential space.** 2 is similarly false, because making a choice that actually matters--that actually has influence or control over something within the play-space--is an event, which can be counted. And 3, while not quite so manifestly false, is still pretty clearly a problem, given the formal discourse on the topic in many fields, all of which I am not fully qualified to directly analyze (philosophy, law, and psychology, maybe economics, maybe more besides.)

Nobody is denying any of those 4.

1. We agree that agency exists.
2. We agree that agency can be counted. HOWEVER, since agency is subjective a countable moment for you may not be a countable moment for me, even if we are in the same moment with the same actions and options. "Matters" is subjective and just because the choice matters to you, doesn't mean that it matters to me.
3. We agree that agency has meaning.
4. This is what we have been saying the entire time. Narrative games offer equivalent agency to traditional games, not more as many people in this thread have claimed.