The Forge Archives

Archive => GNS Model Discussion => Topic started by: james_west on May 09, 2001, 08:45:00 PM

Title: theoretical game spaces (17k in images)
Post by: james_west on May 09, 2001, 08:45:00 PM
I'd not heard of the narrativist/gamist/simulationist axes before looking over this website a few days ago. However, I've looked over not only the material here but related material on other sites, and I've got a few thoughts about the system. I'm going to do my best not to use any mathematical jargon.

It occurs to me in these discussions that it appears that people are defining a theoretical gaming space as appears in Figure 1. In this space a person or game's tendencies are defined by how close they are to each axis; in this example, the person is about 70% gamist, 15% narrativist, and 15% simulationist. Note that in this space, having more of one thing by default means that you have less of others; if you're all the way in the gamist corner, there's no room left to move on the narrativist or simulationist axes, each of which is defined by a line orthogonal to the side opposite the point. This is true because this space is (1) of limited size and (2) the axes are not orthogonal (at right angles.)

Further, many of the discussions seem to center around definitions of each of these; to make a reasonably accurate analogy, a space like this is frequently used to define colors; replace the poles with red, green, and blue. By this analogy, there seems to be a lot of discussion about exactly what wavelength counts as green. I'll return to this later.

Those of you familiar with photoshop will see that essentially what I'm defining here is a color wheel; by defining a location on this figure, you're defining hue.  However, hue isn't all you need to define a color; you also need intensity or brightness. This is because in an absolute sense, increasing the amount of green doesn't decrease the red or blue; it just decreases the red and blue relative to the green. The color space is actually orthogonal, then. If you want a way to plot color which needs no external referents, you're better off using three orthogonal axes. I submit that this is likely to be true of the Narrativist, Gamist, and Simulationist axes as well; increasing one component doesn't necessarily decrease the others (I'm sure you can imagine games that do poorly on ANY of these axes, for instance). In this case, we have a gaming space that looks like Figure 2.

Getting back to my sense about the nature of some of these arguments, consider the following rotation of axes (reducing to two dimensions for clarity).

In this diagram, the star (perhaps a particular game) is at 3 blocks on the narrative axis and two blocks on the simulationist axis. It is clearly more narrativist than simulationist. However, in figure 3b, the axes have been rotated slightly (a slight rotation of axes may correspond to some difference of opinion as to exactly what each axis represents - there have been plenty of discussions in which people propose different definitions of each of the three axes.

Now the star, which hasn't moved, is approximately equivalent in its coordinates on the two axes. The game is no longer narrativist, it's equivalently simulationist and narrativist, even thought the game itself hasn't changed. I think I've pretty much summed up the theoretical underpinnings of a wide variety of arguments I've seen on this (and similar issues) this way.

The next important point, though, is that these two systems are defining EXACTLY THE SAME SPACE. While the location of any given point is different in the two systems, any location in one system can also be given a location in the other system; further, there is 1:1 mapping between the systems.

To see why this is important, let's imagine that when I came upon this discussion group, the claim was that all games were either simulationist or narrativist; it was a dichotomy, rather than a trichotomy. In this theory, then, any game could be placed somewhere on these axes. One might imagine, then, that there could be substantial argument about where to put "poker" on these axes; it isn't really a good fit anywhere. (I suspect people would wind up deciding it was simulationist, since it more clearly isn't narrativist.) However, when we add the gamist axis, where to put it is suddenly very clear; this is a game with basically no simulationist or narrativist elements whatsoever. It is purely gamist.

The problem I have with the gamist/simulationist/narrativist axes is that I suspect there still aren't enough axes (or they're not sufficiently orthogonal, a point I'll address later). Let's take "Candyland" as an example. For those who don't remember it, it's a game in which one moves a token along a predefined path based purely on random elements. It's not a simulation of anything, there are no points at which it's possible to make a decision influencing the outcome, and unless you grossly distort the concept, there's no story being told. This is a game which has pretty much none of these three components. You may argue that it's a pretty lousy game (and I wouldn't argue back too much) but if you think about it, there's still a reason that kids voluntarily play it. Playing any game is a social activity, and socially interacting with other people is fun outside of any other considerations.

I submit that there are a variety of Role Playing Games which fit this model very well. Consider games like Macho Women with Guns or Human Occupied Landfill. These are games in which narrative development is usually cursory at best, it's not really supposed to be an accurate representation of anything, and I seriously doubt that there are any power gamers that play them; nobody plays HOL "to win." I'd call this the "beer & pretzel" axis, or to keep in the more serious vein of the other axes, the "social interaction" axis.

I also claim that this axis is orthogonal to the others, rather than being a simple remapping of the other axes. For instance, computer games have a zero on this axis, but can clearly have strong gamist or simulationist elements, and I submit that, bastardized though they are, computer RPGs have at least some narrativist element. So the "social interaction" axis is distinct. I'm not trying to claim you should change the name of the forum to Gamist/Narrativist/Simulationist/Fostering social interaction; I'm claiming that while the axes you've defined are important, they're clearly not the only axes relevant to defining gaming style, or even goal.

Not that this comes as a surprise to you, I'm sure. There are, of course, a lot of old systems for categorizing players; the one that comes to mind included titles like power gamer/romantic/puzzle-solver/munchkin. These titles, though, fit very well into the G/N/S axes. However, one of the web sites that I read regarding the G/N/S system also included another triplicate: Drama/Fortune/Karma. While the G/N/S axes seek to define the goals of the players (I submit social interaction is an equally valid goal, as discussed above), the D/F/K axes seek to define the methods of conflict resolution. There are many different aspects to gaming; you can come up with one set of vectors for goals (and I suspect that they can be clearly defined into more than three, or even four, vectors), another set for conflict resolution, another set for "stance", etc. Any of these sets of definitions could, ideally, be orthogonal (non-interacting, distinct).

However, I've noticed in discussion that they seem not to be. For instance, while the G/N/S axes pretty clearly are only intended to speak to the goals of the players in playing the game in the first place, there seems to be a strong understanding that narrativist goals are more closely aligned with either the Drama (GM fiat) or Karma (ability score) methods of conflict resolution, and are somewhat distinct from the Fortune (die rolling) method of conflict resolution. I don't think this is neccesarily true; I've frequently found that random events spur the imagination. However, I do think that this is a major source of confusion in discussing these issues. There seems to be a strong tendency to believe that a game's philosophy is entirely defined along the G/N/S axes, and so elements that more perfectly fall into other spaces are projected into this space, frequently resulting in fairly gross distortions.

To summarize my points from above:

(1) Many disagreements seem to arise over mild differences in placement of the measuring axes (closely equivalent to differences in definitions of terms.)

(2) The three axes (G/N/S) used for these measurements may be insufficient to quantify the goals of a game, even if everyone has the same understanding of the definitions.

(3) It's important to keep goals, methods, and other elements of game philosophy separate; I believe that they are separable, and mixing them is bound to lead to a distorted understanding.

(4) I didn't come up with these until I finished writing, and so I may have incorrectly summarized my own article.

Just my $20 worth. Incidentally, I've noticed I seem to be developing a bad habit of my first post in any forum being an attempt to undermine the basic philosophy of the forum itself. You may interpret this in one of two ways: (1) I'm trolling, or (2) I'm trying to more completely understand the philosophy which led to the creation of the forum in the first place.

            - James

Title: theoretical game spaces (17k in images)
Post by: Ron Edwards on May 10, 2001, 06:25:00 AM
Hey James,

Yay, another multivariate dude! Some day we'll discuss elliptical Fourier analysis and factor rotation and make everyone's heads blow up.

First of all, before anyone gets into the "categorizing gamers" issue, I want to clarify that G/N/S applies very nicely to BEHAVIOR. A person in a role-playing situation does X, Y, and Z during that game - I'm reasonably confident the specifics of his behavior can be categorized. (Now, say this person does Gamist behavior very consistently; I now exercise shorthand and call him a "Gamist," and everyone gets mad at me about it. Can we just skip to the point where we're friends again?)

And then, independently, we're looking at an actual RPG, and again, G/N/S applies very nicely to its DESIGN, in terms of what decisions/behavior are facilitated and rewarded by the rules, system, setting, and even things like layout and organization.

Whew. Now to address James' models.

1) I (as many know) am highly skeptical that a person "blends" G/N/S goals in a given instance of role-playing. I am often misunderstood to be saying that a person cannot DISPLAY more than one of them, but this is not true. I am simply skeptical that a given behavior during a role-playing situation can be anything except ONE of the three goals. Observationally, I've found that most role-players do tend toward one of them, across their actions, rather than 1/3 + 1/3 + 1/3 over time. (If you, whoever's reading this, DO display this distribution, fine - I said "most" and I meant it.)

2) Therefore I've never liked the triangle, because I have no idea what the middle represents, if we're talking about a given behavior/instance of role-playing. I like the 3-d graph that James proposes better, in terms of the full independence of the coordinate-combinations, but it still strikes me that the "maximal" coordinate-space (high values on all three axes) is empty.

Then again, if we're talking about a WHOLE person over time, then sure, the middle of the triangle would be one of those "thirdy" people, and correspondingly, the coordinate-space I mention above would be occupied by said people.

3) Rotating the axes - in the analyses I'm familiar with, rotating the axes is carried out SPECIFICALLY when the variables in question (the original, measured variables) are considered INDICATORS of some non-directly-measurable quality in question, and the measured variables themselves are not of interest.

To clarify: rotating axes is standard for measuring "intelligence" (do NOT get me started; we're not discussing the validity of the test, just its methods, OK? if you must, go to private e-mail). Not one of the things being measured on the test "is" intelligence - the tester is looking for a RELATIONSHIP among the answers that is as strong as can be found. Thus, in calculating a regression line through the multiple axes (scores on the questions), one is seeking the most solid line possible.

However, if performance on those questions WERE, in and of itself, considered "intelligence," then rotating the axes becomes an abomination - it forces higher regression coefficients into the analysis than the data actually indicate, resulting in very high "r" values and an inflated sense of certainty about the results.

So my question for James is, what justifies the rotation, in this case? Say we're talking about role-players, not game design. In this case, I think the behaviors of G/N/S are rather easy to spot, and that those behaviors ARE what we're graphing/analyzing - so rotation doesn't seem called for. There's no "intelligence" that sums/multiples from their interaction to consider. What do you think?

3) Humor and parody add a real twist to the situation - such role-playing games cannot be considered in isolation, but in conjunction with what they're making fun of. Thus Hol and Macho Women with Guns, and Paranoia, and Elfs for that matter, are COMMENTS on other games that DO have strong definitions (or specific confusions) using G/N/S. Also, for any of these funny games, they too may be either coherent or "confused" in G/N/S terms, on their own.

Macho Women with Guns, for instance, is making fun of the intense seriousness and background material found in many Simulationist games of the 80s. Furthermore, its own emphasis is unabashedly, balls-to-the-wall Gamist and proud of it. (Personal thought: all the elements of character creation, resolution, and so forth in MWWG seem like an extraction of specific elements of Gamism found in Champions.)

Hol, on the other hand, is making fun of the purse-mouthed, pseudo-academic, meticulously inoffensive settings so commonly found in many Simulationist-ish games from the late 80s and early 90s. However, its system resolution (despite the funny names and text) is pretty much the same as one finds in the very games it mocks, so I consider it to be a mixed success in design terms.

Those are my takes on those examples, anyway ...

Great post, as usual, James - I'm interested in seeing where this one goes.


Title: theoretical game spaces (17k in images)
Post by: Ron Edwards on May 10, 2001, 07:44:00 AM
OK, this is a follow-up, given some musing.

1) G/N/S is not intended to be applied to any games or activities besides role-playing. This is a pretty important point - sometimes, non-role-playing activities are brought in for analogy purposes (I did this recently using tennis), but not as members of the subject matter at hand.

2) Back to rotation, I see that you are getting at the problem of differing measures - if one observer scores (e.g.) game A at "7" for Narrativism, and this other guy scores it at "5," then we have a problem. The axes for each analysis do not correspond.

At this point, we're nowhere near a universal set of measures for these things. G/N/S classification has to be in the "close enough for government" category, I think. Part of the discussion of Simulationism at the moment is going a long way toward bringing everyone's axes closer together.

3) "It's important to keep goals, methods, and other elements of game philosophy separate; I believe that they are separable, and mixing them is bound to lead to a distorted understanding."

I agree. Among the many other elements are character-creation, reward-punishment systems, stance enforcers, D/F/K mechanics, and more. And within some of them are whole worlds of analysis, most especially (1) dissecting Fortune into probability distributions and (2) within Narrativism, the concept of Premise.

I've amassed a huge body of data on all of these things and their history in role-playing games, and one of my main frustrations is getting people past "G/N/S 101" and into a postition when we can actually talk about them.


[ This Message was edited by: Ron Edwards on 2001-05-10 11:49 ]

Title: theoretical game spaces (17k in images)
Post by: james_west on May 10, 2001, 08:54:00 PM

Your "point 2" in the second message is the bulk of what I mean by rotation of axes.  Clearly, the discussion of Feng Shui in this forum shows that people currently have drastically divergent ideas about what each of the definitions means.

I suspect that this is greatly exacerbated due to everyone (including you) tending to forget that this is just supposed to be a method for categorizing goals. For example, you frequently seem to call a particular behavior 'gamist' or 'narrativist', whereas actually at best in this system that behavior can be used as a pointer to the individual's goal in expressing that behavior (if you want a system of categorizing behaviors, that's a horse of a rather different color). While it is true that some methods are probably more amenable to achieving those goals, in RPGs (as in life) methods and goals ought to be held seperate in the mind.

However, there's another reason for rotating axes when doing modeling; that's to see if, through a proper rotation of abstract axes, it's possible to make one of them extraneous. It's frequently true in modeling that you don't know which variables will turn out to be relevant, or even how many axes there are. (This mostly works best if you actually have some sort of concrete data, and you can assign a computer algorith to figure out how many dimensions it takes to represent it.)

As a simple example, let's imagine that we're in the E. Gary Gygax School of Game Design. In this school, the only relevant issue to game design is quantification of character traits. We are told that there are two possible dimensions; character classes, and skills, or combinations of the two. This seems to be a two dimensional system. Thinking about it, a clear third area occurs to us; descriptors. These are non-numerical traits, used exclusively in some games (OTE) but used to supplement many, many systems. In our first pass, then, it seems that our system is three dimensional.

However, upon a little further thought one realizes that character class in D&D is primarily just linked sets of descriptors. D&D, then, is a combination of a few descriptors (class, race, alignment) and a few skills (into which I lump stats, which are just specialized skills) (aside; this is actually a pet peeve of mine; stats and skills are frequently both used to measure the same ability, with conflicting results. This is non-unique mapping, and it's a pain in the *ss in any context.)

So we're back to two dimensions. Now I wonder if I can really reduce it to one dimension; skills on one end, descriptors on the other, and EGG's brilliant new system in the middle. I don't think so, though; I can recall examples of systems with any combination of the two, which implies a genuine multi-dimensional space.

At this point I start wondering whether equipment should be added as a seperate dimension (it's really important in a lot of games ...), or is really a combination of numbers and descriptors.

You see how this works ? Knowing how many dimensions your space has is at least as important as precisely defining the orientation of the axes you're going to use to map it.

On a different topic, the fact that the grouping of HOL and MWWG was clear to you, and you even added a third example (which I'd almost put in my post) ought to be suspicious to you.  These are games that clearly have some strong element in common, which fits badly on the G/N/S axes (or, really, any of the other spaces discussed; stance, resolution, etc.)

Finally, I think there's no particular reason to limit this set of goals to canonical RPGs; they work very well with boardgames (as a Gordon Landis mentioned in another thread) and many other games. So limiting them also tends to marginalize many games that have strong role-playing elements but are not clearly within the RPG paradigm.

                     - James

[ This Message was edited by: james_west on 2001-05-11 00:57 ]

[ This Message was edited by: james_west on 2001-05-11 01:01 ]