The Hidden Bartle Type

Often when I write about games, I'm interested in an experiential perspective on play and in the kinds of social and political structures that evolve within virutal worlds, but TN readers probably have noticed that I'm equally drawn to questions about design processes, about the authorship of digital games.

I think this is why Metaplace excites me so much. It wasn't very long ago when I was last speculated about author tools here, and now here's an ambitious project to create a very powerful example of such a tool.

An attraction to questions of design is a very old part of the culture of virtual worlds and online games. In play within worlds and in the forums, guild chats, and other systems of communication that surround them, there is an identifiable type of player who has strong opinions about design questions, sometimes informed opinions, sometimes malformed one, but opinions that go well beyond "U NERFED MY CLASS!!!!!". Where the attraction to design is a part of the experience of play, and where the player's activities within the game are at least partially aimed at a kind of pure understanding of how the game or world functions (rather than an understanding which is aimed at maximizing achievement).

It's always seemed to me that this approach to play was distinctive enough that it could easily be called a fifth Bartle-type to go alongside achiever, killer, explorer and socializer. Call it subcreator, or if you want to get fancy, demiurge. From MUDs to WoW, there have been games which reward the subcreator within the activity of play in some respect. Some examples: granting creator privileges, bringing the subcreator inside some privileged discussion about live management (various volunteer programs, for example), by allowing for a class of artifacts within the world that were created by players (books in Asheron's Call 1, or the entirety of Second Life). But with the exception of Second Life and maybe Neverwinter Nights, there has been less and less out there for the subcreator types to do. So for this reason alone, I'm rooting for Metaplace. Not because player-created content is superior or because it magically resolves some of the problems of existing virtual world designs, but simply because it will have tools for a type of player that normally is exiled to the MMO equivalent of the Island of Misfit Toys.


Comments on The Hidden Bartle Type:

Tom H. says:

I think the reason I found A Tale in the Desert so compelling was because it catered to the demiurge. I had to quit when I realized I'd been destroying my sleep schedule in an attempt to reverse-engineer the mechanism behind mushroom growth, which spanned 2 out of every 7.5 hours. But only a couple of thousand people agree with me, nowhere near mass market - how could you bridge the gulf between ATitD and the subcreatorish fanbase I'd think you'd find in the Sims?

Posted Oct 10, 2007 3:43:11 PM | link

Bart Stewart says:

Since I've been involved with this area I've felt that design -- the creation of systems -- is one of the expressive modes of the Explorer type. I see design as a process that Explorers use for increasing their understanding of how a system works.

Of course that doesn't mean other types can't be creative. But they seem less often to be creative in the sense being asked, of design creativity, than the Explorers.

That said, there's a sense in which we might say there's a fifth player type (or a ninth, following Richard's new model from Designing Virtual Worlds). This would be the type who's transcended all the others, who's experienced and understands and appreciates all four/eight types, and who operates consciously using the strengths of all the types in a way that's balanced over the long term.

Such a player might be able to do a great job of creative design, but I think I'd say that he or she was tapping primarily into an Explorer strength in doing so.

All of which is obviously debatable; I'm just offering another viewpoint on an interesting question.

Posted Oct 10, 2007 4:41:01 PM | link

ErikC says:

"Where the attraction to design is a part of the experience of play, and where the player's activities within the game are at least partially aimed at a kind of pure understanding of how the game or world functions (rather than an understanding which is aimed at maximizing achievement)."
Exactly!

Bart would this fifth type be a sort of creative guardian?

Posted Oct 10, 2007 7:11:08 PM | link

Richard Bartle says:

The thing about the player types model is that it's a model. It's not a list - you can't just add a new type - it's a system. There are four types (or eight types in the newer version) for a reason to do with the structure of the model, namely the axes of the graph.

If you identify a new type, then this either means the model is wrong/incomplete, or that the type is orthogonal to the model, or that the model doesn't apply to the type.

For example, gold farmers don't play for fun. My types model only explains why people who play for fun play, therefore it doesn't apply to gold farmers (or many journalists, academics and customer service staff).

Also for example, people have noticed that one of the reasons players give for playing is "to become immersed", which, while not associating with any particular player type, does match very well the process of changing type as players progress along their hero's journey (that is, the more you feel that the character you're playing is you, the more immersed you are; your player type reflects your current motivation for playing, but the overall aim of greater immersion is the result of this).

So, where do builders fit in?

Well, if they do indeed fit in, it would depend on why they're building. If people are building in order to make money (if that's how the world measures success) or show off their building skills, that might be an achiever-like motivation; if they were doing it to see what kinds of things can be built and to push the building system to its limits, they'd probably be explorers; if they were doing it to enhance their world and make it more attractive or salubrious for others, they could be doing it for socialiser reasons; if they just want to crash the grid or spread a plague, that's killer/griefer behaviour.

Now it seems to me that you're not actually talking about any of these kinds of building. You're talking about people who build because they want to design. They still get some fun from playing, but they're transitioning from wanting to play to wanting to design.

OK, well this wouldn't be a fifth type in my model, it just doesn't fit. However, as with immersion, it's enough about fun and play that it ought to link in somewhere - it's not as if you're suggesting we add a fifth type for "people who regularly wear hats in their daily life" or anything equally irrelevant. Some players really do like creating for reasons that are to do with creating, not to do with playing for regular fun. Simply adding yet another dimension to the player types model isn't a useful solution, though, as it doesn't explain anything, it just attempts to describe it. A model that doesn't explain isn't a model.

As it happens, I've given this some thought myself in the past. I haven't written anything up on it, because it's not fully fleshed-out, however I hinted at it in my AGC keynote a couple of years ago. Look at the second slide on page 19 here. What I'm saying is that there are designer types as well as player types, and they follow their own hero's journey; they use their designs to articulate aspects of themselves that they couldn't say any other way. In the same way that there's a "main sequence" of player progression, there's one for designer progression: they start off as players (experiencing content), then become level designers (implementing form), then progress to being fully-fledged designers (implementing experience), then graduate to creative director (experiencing form). At this last stage, they can read the designs of other designers very well; this is why many top designers don't play games all the way through, they only play enough that they can grok them.

More pretentiously, this doesn't just apply to virtual world design, it can apply to any art. I remember being struck by how similar Scott McCloud's description of the various stages of comic artist fitted in with this, for example.

Anyway, to come back to your original point: the "fifth type" as you describe it concerns the point where the player begins the switch from being a player to being a designer. I'd expect it to happen to high-end explorers (what I call hackers in my 8-type model) and socialisers (what I call friends), but it could happen to anyone sufficiently inspired. They start to move away from playing for regular fun and start to play for designer fun. The more they move in that direction, the less they'll need to play and the more they'll need to design, which for some may be giving up too much, but for others may be opening up parts of their self that have been bursting to break free for ages.

That's my take on it, anyway. Feel free to disagree!

Richard

Posted Oct 11, 2007 4:31:14 AM | link

Mavis says:

Thank you - I now know who I am.....

Posted Oct 11, 2007 7:00:22 AM | link

C-Park says:

When Richard talked about the stages of art mastery, those stages spoke more to the path people take through an emotional space than to the dimensions of the space itself. Richard's player/world - acting/interacting space for the player-for-fun is not the one traveled in the artist's journey from learning the mechanics of art to fully appreciating the artistic decisions of others. I'm sure Richard has given thought to what the dimensions of that terrain are. If he feels the ideas are far enough along I hope he can share them with us. Creativity and inner-directedness come to mind, but others in this forum could certainly offer principled alternatives.

We've been considering game design as the creative activity within games, but that isn't the only form of creation we could enable. If there is an artist's journey we can facilitate, mechanisms supporting it might well be generalized to a broader range of artistic endeavors than game design -- however tempting it might be to think of them all as "content".

Arranging furniture in Star Wars Galaxies, A Tale in the Desert's art challenges, and development of a battle plan in EVE are all forms of artistic expression supported to some extent by game mechanics. The out-of-game home shows of SWG, voting on artworks in ATITD, and EVE's Command-And-Control skills are all mechanisms that help identify, nurture and develop players' artistic bents. Still, they do not seem to be engineered to that end.

Just as Richard developed dimensions along which play-for-fun players could be placed, measurable player characteristics that help chart a path for the artist's journey would provide an important tool for game designers.

Posted Oct 11, 2007 12:12:49 PM | link

Dmitri Williams says:

I'm partial to Nick's 3-factor model of motivation: achievement, socializing and immersion. The reason why is I've seen it now tested and validated twice on large samples and it's mathematically solid. I like the intuitive value of Richard's 4-part model (especially its narrative), but it doesn't hold up as well when you do the math and testing on a group.

Having said all that, Nick's model doesn't have a creative component, and that's an interesting suggestion. Maybe it'd also factor out and then we could see how prevalent that player type is.

Posted Oct 11, 2007 2:00:33 PM | link

Timothy Burke says:

Thanks, Richard. That's very useful. C-Park's comments seem very to the point as well. He reminds me, for example, of the person I remember running into in the early history of SWG who had done this amazing interior design work on his house using fishtanks. (Early on to get some stuff onto walls, etc., you sometimes had to put it on a table and then use commands to move it horizontally across a room through empty space--there were a lot of creative kludges that people invented.) What that player was doing was fun, it was a form of play, and it struck me as a foundational attitude towards play, a motivational "type" that was parallel to socializing, exploring and so on but wasn't implicit within those types.

Posted Oct 11, 2007 2:01:03 PM | link

Richard Bartle says:

C-Park>When Richard talked about the stages of art mastery, those stages spoke more to the path people take through an emotional space than to the dimensions of the space itself.

Yes, it's not the same domain that player-fun takes place in at all.

>I'm sure Richard has given thought to what the dimensions of that terrain are.

I have, but I don't know enough about art theory to be able to state them with any confidence. The ones I had in my AGC talk were experiencing/implementing versus form/content. The consumers of the art (players in our case) experience the content.

>If there is an artist's journey we can facilitate, mechanisms supporting it might well be generalized to a broader range of artistic endeavors than game design

One of the frustrations I have regarding "art and computer games" is that whereas artists are all too ready to consider art created within computer games as art (eg. Machinima), they are often reluctant to call computer game design itself art. So yes, a game that helped lead people along an artist's journey could be something special, but I'd still like to see game design itself getting some recognition as an art rather than a craft.

Richard

Posted Oct 12, 2007 9:38:42 AM | link

Richard Bartle says:

Dmitri>I'm partial to Nick's 3-factor model of motivation: achievement, socializing and immersion.

Well, it's not exactly a model: it's a categorisation.

>I like the intuitive value of Richard's 4-part model (especially its narrative), but it doesn't hold up as well when you do the math and testing on a group.

See, this baffles me, because I look at this achievement/socialising/immersion categorisation and it appears to me as if it supports my player types model. It omits griefers and explorers, which are generally smaller populations anyway, and it identifies immersion as a driver.

That said, if this three-types approach is as rock solid as you say, it means the concept of a griefer is merely an urban legend, and it means you can't become immersed through achievement or socialising.

>Having said all that, Nick's model doesn't have a creative component

In the past, my player types model has been criticised by some of the MOO crowd for not having "builders" as a category. The reason they wanted it was because that's what a lot of people in those virtual worlds did. However, it's a mistake to associate what people do with why they do it; in a virtual world where the gameplay concerns building, then you'll get the same player types as in other virtual worlds, except expressed through building rather than engaging in combat.

Timothy's question opened up the possibility that the act of building was in and of itself a motivation. To fit this into my model, I have to hypothesise an artist's journey, and put the players who do this as being on the cusp of switching from wanting to experience content to wanting to implement form. An analysis of Nick's data might be able to identify the existence of some "creator" class, but it couldn't say anything more than this because it's not a model - it doesn't attempt to say why things are the way they are, just that they are.

Richard

Posted Oct 12, 2007 10:04:08 AM | link

Dmitri Williams says:

You're right, Richard. Your model and his scales have different goals. And I think the discussion here suggests that there could be factors added to his scales to see how prevalent griefing or creating are. There's nothing about his scales that says they are all-inclusive. They're just good at identifying the three components he's teased out so far, and there could certainly be others.

The thing that I have a hard time getting past with your 4-part categorisation is that if you say it's not a model then what's its use? Ultimately, the thing has to be testable. If it isn't, it can't be falsified or confirmed and isn't of any true use. It can be a thought exercise, but if it turns out to be inaccurate, then it's worse than nothing.

Nick's more familiar with the testing phase, so maybe he can weigh in on whether the components in your categorisation stood up when subjected to factor analysis of questions matched up to the four categories.

Posted Oct 12, 2007 11:54:58 AM | link

Thomas Malaby says:

I don't understand your question, Dmitri. As far as I can tell, Richard was saying that his scheme *is* a model, and that Nick's isn't. In any event, this may be unprofitable term-wrangling. It seems to me that Nick's and Richard's are both "models" if we mean that in the *heuristic* sense. The usefulness of either scheme lies in the degree to which it contributes to reliable claims about what virtual world participants do, rather than in some grander claim that the models are real themselves. (Perhaps Richard is making this kind of grander claim for his scheme? I suppose that might be the difference here.)

In the sciences (speaking broadly), reliable claims can be generated through testing, of course, but also through exploratory work, or even critical evaluation (like reading an x-ray), right? Sure, some kind of inquiries are better suited to hypothesis-tessting, and others to critical inquiry, etc, but if we make the mistake of thinking that the testing is utterly different from other empirical inquiries, we'll really set a hard but false divide not between parts of the academy (humanities, natural sciences, social sciences), or between specific fields, but even between different scientists in one field (field scientists vs. lab scientists, say), or maybe between the parts of *one* scientists' work practice. Experiments aren't not where "true" claims exclusively lie.

Posted Oct 12, 2007 12:15:25 PM | link

greglas says:

Just because it seems rather relevant -- on the left sidebar there's a link to Damion talking about ways to use and misuse the Bartle types. link.

Posted Oct 12, 2007 12:37:03 PM | link

Thomas Malaby says:

Erk. Last sentence should start "Experiments are not..."

Posted Oct 12, 2007 12:39:28 PM | link

greglas says:

Oh, and Richard's comment there, and his agreement with Damion's comments in the post, give a pretty extensive take on what he sees as the limitations of the model. Better than anything I can add to this.

Posted Oct 12, 2007 12:44:45 PM | link

Thomas Malaby says:

Thanks, Greg, Richard's comments there clear up one confusion for me: for Richard, his scheme is a model (and Nick's not) because of its internal coherence and comprehensiveness -- an underlying set of ideas generate the types they have and no more.

Posted Oct 12, 2007 1:28:46 PM | link

Timothy Burke says:

I don't think typologies have to be testable, in fact, to see them as something that needs to survive rigorous testing is to miss the point of typology in the first place. Typologies are heuristics; they're ways to organize complexity into manageable units for the sake of discussion. They're necessarily provisional.

Look at the mother of all typologies: Linnaean taxonomy. In some sense, that can't be "rigorously tested"--there are still endless debates and always will be such between biologists about where to classify some organisms, and that's with very good empirical data about genetics and behavior, often.

Posted Oct 12, 2007 1:58:00 PM | link

greglas says:

Right. Generally I think the typologies say as much about the person making the classification (what constitutes a "worthy" line of division) than it does about the object of study.

A.k.a. Stealing from Borges, animals might be divided into the following types:
1. those that belong to the Emperor,
2. embalmed ones,
3. those that are trained,
4. suckling pigs,
5. mermaids,
6. fabulous ones,
7. stray dogs,
8. those included in the present classification,
9. those that tremble as if they were mad,
10. innumerable ones,
11. those drawn with a very fine camelhair brush,
12. others,
13. those that have just broken a flower vase,
14. those that from a long way off look like flies.

Posted Oct 12, 2007 2:07:35 PM | link

Dmitri Williams says:

I disagree, Tim. If a typology isn't testable (and tested), it's ultimately navel-gazing. What is the point? Thought exercises are fine, but they eventually have to leave the nest and be put to use. If the Linnean taxonomy hasn't been tested, that's a problem with a lack of data, not the approach. The point is that people are constantly collecting data and going back to the taxonomy and refining it so that one day it might actually be totally usable. I am no biologist, but I am sure that it's not just there for fun or heuristic value. It's there for its use in the field, for testing empirical objects and using them to construct and refine larger sets of theory.

The problem with your reasoning shows up in Damion's thread. People will use the typology, having assumed that it has been validated in some way. And they then run the risk of getting things wrong. If the thing had been tested ahead of time and found valid, it wouldn't be the cause of things gone wrong. So, caveat user.

I think Richard's typology--and it isn't a model because models are intended to be tested by their nature--is very useful as a way of thinking through motivations and play styles. With the notes in that post, it looks pretty darn comprehensive, too. But that doesn't mean it's held up to testing, and in both empirical science and in game development, that's where the rubber meets the road. Are those concepts really at play? Do they overlap and how much? How common are they? Those are the things that validity tests answer.

Nick's typology--although clearly not comprehensive--has held up to testing and therefore can be used. It can and should be continually refined to see if creators, griefers, etc. stand up as motivations as well. The OP is a great argument that it's not comprehensive.

BTW, as an aside, Nick's typology is about motivations, while Richard's is a larger typology of both motivations and actual actions within worlds. Thus, Nick's is more limited.

Thomas, I make no distinction here between kinds of empirical testing. If there was a critical, ethnographic or some other qualitative empirical testing of the approaches that could speak to their validity and generalizability, that'd be great. I haven't seen one, but maybe it's out there.

Posted Oct 12, 2007 2:56:49 PM | link

C-Park says:

Agreeing with Dmitri... Richard's and Nick's work is different and should not be directly compared.

To Thomas's point ( and disagreeing with Timothy! ), I don't believe that for us the test of a succesful model can be measured by it's comprehensiveness, internal coherence or even descriptive accuracy. If our underlying goal is to provide tools for game designers to implement design decisions -- like helping players move through some psychological arc or catering to their underlying motivations -- then the space through which they move has to be defined with dimensions that:
o inform the designer's decisions on mechanisms by which players can supported or directed
o allow a player's position on them to be measured in-game.
It is in that first objective that I think Richard's model is less comprehensive.

His dimensions of player/world - acting/interacting don't give much insight into why players move from one place to another in the space. Designers must apply experience and intuition to craft game mechanisms to accomplish their objectives. After that, using Richard's formulation, good solid design lessons can be learned about how to transition players from one state to another. How? Measure, remeasure, compare, and see what game mechanisms were used. Even accepting that we don't have such complete instrumentation of any particular game now, at least in principle the measurements and analysis could be taken given the dimensions he defines. As Richard's already pointed out, there are some things the game designer might want to say that his dimensions don't express. And as Dmitri's suggests, Richard's dimensions might not stand up to empirical testing. That's where Nick's research provides a complimentary effort.

His taxonomy gives insight into players' self-reported motivation, but his methodology does not provide in-game measures to test game mechanisms. Additionally, the player-reported motives of socialization, achievement and immersion seem open to further refinement both to better inform game design and to make them more measurable. Perhaps achievement can be factored into accumulation and creation? Perhaps immersion can be exposed as either control or something that underlies real-world frustration? I know Nick must be devising experimental tools to uncover those dimensions as we speak! :)

Nick's (and others) research can help define a taxonomy of dimensions we can use. They in turn can define the spaces in which our players' motivations live. We can devise and test tools to craft player experiences in those spaces and measure their success. I can understand Richard's frustration with getting game design accepted as an art. But I don't think we're at the point of painting the chapel -- we're just learning how to mix the paint, stretch the canvass and hold the brush!

Posted Oct 12, 2007 3:24:41 PM | link

greglas says:

>> But I don't think we're at the point of painting the chapel -- we're just learning how to mix the paint, stretch the canvass and hold the brush!

The inability to stretch canvas is no reason wait for scientists to tell you how to make art.

Posted Oct 12, 2007 3:55:35 PM | link

greglas says:

And ever so oddly, I think that comment relates pretty well to the opening post.

Posted Oct 12, 2007 3:57:09 PM | link

Thomas Malaby says:

But Dmitri, you can't on one hand say that it's either testing or it's navel-gazing (something I've heard elsewhere around TN at times) and then acknowledge that other kinds of empirical inquiry (critical expert judgment or exploratory observation) can produce reliable claims. In fact, of course, they're all involved to greater and lesser degrees in every attempt to generate grounds for a reliable claim. This is part of a broad tendency that one sees to divide inquiry such that empiricism=testing and anything else is opinion. It just doesn't work that way. I know we've agreed on that in the past, so I guess I'm just making sure that the language of "testing" or "falsification" or "validation" doesn't creep in and do the damage it normally does; i.e, turn science from being something that is always about *provisionality* of its claims into something about what is "true/not true".

There's an example of that creeping in when you say, "If the thing had been tested ahead of time and found valid, it wouldn't be the cause of things gone wrong." The problem here is that this assumes that once testing has validated something then its status as true is unassailable. This is not how science (in fact, all empirical inquiry) works, in my opinion. Sure, some models have been so supported for so long by different kinds of application (not just testing!) that a single contrary outcome is unlikely to dislodge its claims to reliability, but the point must always be that it is up for revision nonetheless.

Tim (or Greg) should pipe up if they think I'm misrepresenting them, but I take the above to be what they're saying as well.

Posted Oct 12, 2007 4:26:02 PM | link

Dmitri Williams says:

No, there's no ultimate proof and everything is always up for revision and refinement. I'm with you there. My point is that there needs to be some checking or it's all talk.

Posted Oct 12, 2007 4:30:20 PM | link

Timothy Burke says:

Checking in the sense of "ways to see whether this is a good description of both observed reality and underlying behaviors"? Yes. But if you say, "This can't be tested in some final and absolutely rigorous way, throw it out", that's an over-the-top kind of positivism. Big excluded middle there that's especially relevant to typology and model-making.

Not the least of which because models, no matter how tested, aren't reality. That's why they're called models. No model holds up to the staggering complexity of lived human reality, even when we bound that to something like "behavior in virtual worlds". Models are useful because they simplify, reduce, truncate, isolate something about that complexity, and allow predictions to be made and tested from that simplification. But this means that models and typologies are at least partially a matter of art as well as positivism: a typology is usefully repeated when its proposed reduction strikes a nerve, evokes an experience, feels right. This doesn't mean that something that feels right can't be tested and found wanting. If we already understood everything about our own experiences, we'd hardly need social scientists at all. But I think I'd mellow out on what "testing" means in this context.

Posted Oct 12, 2007 4:48:47 PM | link

C-Park says:

greglas says:

>>The inability to stretch canvas is no reason wait for scientists to tell you how to make art.

I don't particularly like the idea that art is constrained by technology either. Indeed, wonderful art can arise from primitive tools. But the Venus of Willendorf was chipped from the limestone with a tool, the cave paintings at Lascaux may be 17,000 years old but someone learned to crush the berries to make the pigment, and whatever the reason for Stonehenge those rocks didn't just hop up there by themselves. Without tools those artistic visions would not have been realized.

I'm with Timothy when he says "Models are useful because they simplify, reduce, truncate, isolate something about that complexity, and allow predictions to be made and tested from that simplification." To that end, our artistic vision should inform our tool creation. And since we're breaking new ground we need to make sensible judgements about how well those tools work. Moreover, the players of our games are an integral part of our artistic vision. We should take special care choosing the tools that guide them.

To the extent that the MMOG game designer's art is expressed in the shared experience and journeys of the players, the tools we're talking about for managing players' movements though emotional space are one of the few things that allow us to transcend the vagaries of OS platforms and display hardware. If we don't master those tools much of our "Art" will be hard pressed to make it past the next console fad.

Posted Oct 12, 2007 5:52:00 PM | link

greglas says:

C-Park, I'm with you on all that -- I'm all for better tools.

Posted Oct 12, 2007 6:36:04 PM | link

Dmitri Williams says:

Well said indeed.

Posted Oct 12, 2007 6:59:04 PM | link

Thomas Malaby says:

I'm still wondering what your response is more generally to Tim's and my comments, Dmitri.

You wrote:

The point is that people are constantly collecting data and going back to the [Linnaean] taxonomy and refining it so that one day it might actually be totally usable. I am no biologist, but I am sure that it's not just there for fun or heuristic value. It's there for its use in the field, for testing empirical objects and using them to construct and refine larger sets of theory.

But this is exactly what heuristic *means*; that it is already being used and thereby continually adjusted and refined to be more useful, but with its users never losing sight of the fact that it's not *real*. Its validity is not waiting for an experiment (afaik, one cannot be done to validate or invalidate the Linnaean taxonomy tout court) -- it is already a reliable model (as is natural selection, for example).

Posted Oct 12, 2007 8:56:59 PM | link

Dmitri Williams says:

I have no issue with a heuristic that actually gets used, tested and then refined over time. I simply am not interested in heuristics that will never be tested or are somehow above testing.

Posted Oct 12, 2007 10:59:23 PM | link

Thomas Malaby says:

Okay, but I'll second Tim's comment that it's a bit problematic to lump all the ways that models are used both to inform research and themselves be revised through that empirical work under the term "testing". Exploratory work (such as used in studies of river formation [as I heard a nice paper about recently]) is not a test in the hypothesis-testing sense. An expert reading an x-ray is not performing a test in that sense either. But both of them are empirical, and they are just as valid of means through which further support or grounds for revision of the model can be generated. Would you agree with that?

And hey, what's up, Dmitri? Do they revoke your quant-sci secret decoder ring if you get caught calling that stuff something other than testing? ;)

Posted Oct 13, 2007 12:43:50 AM | link

Richard Bartle says:

Dmitri>People will use the typology, having assumed that it has been validated in some way.

It has been validated in informal ways (see the earlier Zen of Design post about the GoPets experience, for example). It hasn't been validated in formal ways, though neither has it been invalidated.

The thing is, at the moment it's all we have. If people hold off using it because it hasn't been validated, what are they going to use instead? Are they just going to run blind, or do they have some other theory to inform their decisions? If they do have such another theory, can the rest of us see it please?

>I think Richard's typology--and it isn't a model because models are intended to be tested by their nature

Whoah! It is a model, because it's intended to be used predictively. If you do this, you get that. If the model says that doing this will get you that and it doesn't, then the model is wrong; either it can be repaired, or it can't be and we get a replacement model as a result. Either way, we wind up with a greater understanding of why people play virtual worlds, so we can design better ones accordingly.

>Nick's typology--although clearly not comprehensive--has held up to testing and therefore can be used.

Used how, for what?

My model is intended for virtual world designers to use. It's not necessarily going to be what social scientists want to see, or indeed what players want to see. Nick's typology as it stands is not going to give designers much food for thought, but it could be perfect for, say, investigating how real-world demographics map to virtual-world demographics in order to gain insights into how best to pitch in-world advertisements: a real application using real money giving real results.

>BTW, as an aside, Nick's typology is about motivations, while Richard's is a larger typology of both motivations and actual actions within worlds. Thus, Nick's is more limited.

My player types model concerns what people find fun in virtual worlds. As fun is a motivator, you can infer that the player types map to motivations, but strictly speaking that's not what the model claims. Also, although the actual actions can be used to illustrate how players of a particular type behave, they do behave in other ways, and players of other types may sometimes do the same thing.

I agree that formal tests to support or invalidate my model haven't been conclusive either way. Already, it has been extended to cover informal observations made about the original version; sooner or later, something concrete is going to come along that will break it. I expect this - I want it to happen! As I said in the Zen of Design post, though, if people are going to haul it down I'd kinda like to see something better go up in its place.

Richard

Posted Oct 13, 2007 7:39:30 AM | link

Richard Bartle says:

C-Park>His dimensions of player/world - acting/interacting don't give much insight into why players move from one place to another in the space.

That's right, this was an early criticism of the model, which eventually led to my adding an extra dimension to give 8 types. Looking at the common paths that players took through those 8 types eventually reduced to a mapping onto the middle section of the Hero's Journey. This finally gave the reason why players move from one area to another on the graph: they're following a hero's journey. In the language players use themselves, they're striving to become more immersed.

Richard

Posted Oct 13, 2007 7:49:23 AM | link

Ola Fosheim Grøstad says:

A theory about motivation for eating:

1. some people prefer sweet things
2. some people prefer non-fat food
3. the rest just do stuff

What a perfect taxonony, of course it is rather useless as it misses the main motivation for eating...

This is a statement about player-type taxonomies...

Posted Oct 13, 2007 8:55:53 AM | link

Andy Havens says:

@Ola: Taxonomies describe behavior. Motivation analysis describes underlying causes. You use the former, in many cases, to help qualify that latter.

IE... if a player exhibits behavior qualities of a "killer," he's probably not doing it because he's motivated to make money.

Now... if you want to talk about the motivations that drive all players to play games... that's a different story. We all eat to survive; stop eating, and you quickly become disorganized cellularly and die. Deep rotting stench and decay quickly follow.

What is "the main motivation" for gaming?

I'd say it depends on where you fall in the taxonomy ;-)

Posted Oct 14, 2007 12:52:14 PM | link

Ola Fosheim Grøstad says:

Andy: "We all eat to survive"

HAH! That's a matter of perspective, and one can't get to that knowledge by using surveys or even in-depth psychoanalysis.

Want to discuss player motivations? Walk the narrow path and stick to thick descriptions (or your personal experiences, of course).

Posted Oct 15, 2007 6:21:47 AM | link

Ola Fosheim Grøstad says:

Just another comment, Andy.

What is "the main motivation" for gaming?

I'd say it depends on where you fall in the taxonomy ;-)

Hmm. Yes... That's the problem. Currently "the taxonomy" appears to be the key properties of MUDs that the designers actually aimed for (chat, points, interaction and a fictional world). So what you get is the dimensions of the design-spaces applied to it's users. You could do that with any object we use. Like the motivation for being involved with cars: transportation, speed, coolness etc. People get attracted to the key outstanding properties of the object. People drop the object if the key properties don't fit them. So, such taxonomies are by themselves useless, or rather, no more useful than a description of the object.

If you use the taxonomy to say something about the relationships between the entities in the categories and their surroundings you get a model about something, yes. It might be useful, but it would still be one perspective out of a million (or more), so you also have to argue that this particular perspective is actually a useful one, or at least more promising than the other possible perspectives that were considered...

A useful taxonomy for player types would have to take into account the premises for entrance, why players stay, leave and make transitions (on multiple levels). How we relate to eating is very complex, understanding VWs is too, because we don't really need them, so when people do need them they need them for very different reasons, and the reasons changes over time. E.g. you might start using a VW because you are lonely, escapism, but that doesn't mean you socialize (or admit your loneliness to yourself). You might stay for another reason, make transitions for yet another and exit for yet another.

I sincerely doubt that the needs people get covered in virtual worlds are specific to virtual worlds, though. The key motivation for long-term use is probably ID, but building ones identity is a never-ending process. If you can figure out how hard core users use VW to displace the other means which they otherwise have used to feed their id you might have a more general theory about that specific group. Perhaps.

Still, even with such a theory, how can you use that to improve the design of virtual worlds?

If "the hero's journey" has merits it could be used, but I doubt simple generic theories can do the psychology of an individual user justice. *points to users not needing VWs, thus coming to them with a wide array of needs*

Posted Oct 15, 2007 7:10:15 AM | link

Andy Havens says:

Ola: First... Eating to survive isn't a matter of perspective. Eating to do *more* than survive is. Man may not live by bread alone, but without it (or something like it), he don't live at all. Go long enough without some kind of food, you die. The absolute *need* to eat will color many issues related to food choice. In the history of man, there has never been even one person who could live without food.

However, there has never been one person who couldn't live without any particular game, no matter how unpleasant it might seem. So comparing taxonomies of food preference to gameplay preference is, I think, less than useful.

As is looking for a system that will explain every individual behavior of ever particular individual, which is what it sounds like you are looking for. There are many types and depths of categorical systems. They are useful inasmuch as any person finds them useful for a particular use (redundancy intended), not in order to build a map of the world its actual size.

If I separate my customers (of any service) into four categories based on some kinds of behaviors that link, I observe, to issues I wish to influence, and, over time, I see that doing so helps me design improvements, campaigns, etc... then the taxonomy works. Not because it will predict (or explain) every facet of their existence, but because it is useful to me in a specific way.

Richard's system may be very useful if it describes, in broad strokes even, the motivations and behaviors of even (I'm making this up) half the players in MMOs. If we want to make a game that's attractive to people who like exploration, and look at the parts of a game (quests, new levels, UI, maps, etc.) that are part of the exploration process, we can observe changes to that category.

As an individual person, I may care that a buddy of mine started playing because his wife left him and he had a bunch of free time on his hands. As a marketing person, I care why he makes the in-game choices he does and why he stays or goes.

As you say, motivation for use of games and other pastimes may be very similar or the same. But the role of designers and publishers of games isn't to fathom the hidden depths of the psyche (at least, that's not all they should be doing). And the usefulness of categories that provide a way to helpfully address game situations and improvements isn't circumscribed by a lack of infinitude.

Posted Oct 15, 2007 8:59:20 AM | link

Ola Fosheim Grøstad says:

My point is that the motivation for even something as basic as eating is complex with multiple psychological and cultural layers. We aren't motivated by the threat of death. We just feel like eating particular types of food for a wide array of relatively complex and personal reasons. I would expect the motivations in VWs to be at least as complex, and I believe I've seen/heard enough to confirm that.

The Hero's Journey seems to be more of an aesthetical theory than a psychological one. Perhaps some users of some virtual worlds follow this aesthetic, which in turn begs the question of the chicken and the egg. I can't view it as as a general theory about motivation in VWs.

Btw, we don't need a theory about virtual worlds to tell us that people like specific activities that they also like in general, outside virtual
worlds (winning, tourism, pubs... etc).

If one can claim that a lack of on particular activity (e.g. tourism) hampers the virtual world as a whole, then you'll have something. I don't believe that has been established?

Posted Oct 15, 2007 10:52:13 AM | link

Dmitri Williams says:

Sure, Thomas, I'd call your example testing as well. I was trying to find a generic term that applies to all empirical methods equally.

Richard, you ask if an unverified system is better than nothing, and I'd have to answer "Who knows?"

Nick's three motivation factors can be tested on systems. So, we know that we could ask a population of a game his 10-question series and we'd then know who and how many of them are motivated by social, achievement or immersion factors. If we also tied those measures to what they do or how they played, that'd be a pretty damn useful planning or testing tool. And, like I said before, it'd be more useful still if it added (i.e. added and then validated) the creative element Tim suggested.

And maybe Nick can weigh in here, but I was under the impression that he'd checked on the killer and other factors and found they hadn't held up. Maybe I'm wrong about that. The way to do this is to come up with a series of questions that line up with each of Richard's types. Then you ask them and see first if they cluster together as answer groups. If that works and the math checks out, you then have a tool for measuring how many people fall into these categories and who overlaps. And that's a working model that's very useful.

Posted Oct 15, 2007 12:26:37 PM | link

Thomas Malaby says:

Glad to hear, Dmitri. Hmmm, given "testing"'s clear limitations, I'll leave it to Tim to suggest a better term -- he's much cleverer with that stuff than me.

As for your account here -- I think you're right that you could do all of this, but I think it's important to note that unless you're making the mistake of taking people's representations of their motivations at face value, the qualitative research into their practice (as you rightly put it, "what they do or how they play") isn't just a nice addition -- it's absolutely crucial. After all, at times we know that how they play will contradict their representations. This tendency to take these accounts at face value was one of the limitations of Nick's work, in my opinion.

Posted Oct 15, 2007 1:05:21 PM | link

Dmitri Williams says:

As always, combined methods FTW. But you'd see the value in seeing the baseline numbers for the whole population, right, or a subgroup (e.g. PvP players, raiders, mixes of the two, etc.), right? Like we find that X% of people are motivated for social reasons, and X% are for immersion and X% are for achievement (they aren't mutually exclusive), the people who are high on social and achievement tend to go do activity Y, etc., etc.

Posted Oct 15, 2007 3:27:25 PM | link

Thomas Malaby says:

Close :). In my opinion, you'd find that "X% of people [claim to be] motivated for social reasons, and X% [claim to be] for immersion and X% [claim to be] for achievement (they aren't mutually exclusive), the people who [are more likely to represent why they do it as] social and achievement tend to go do activity Y, etc., etc."

:)

Posted Oct 15, 2007 6:13:54 PM | link

Dmitri Williams says:

We don't trust people, either. That's why we ask several questions that get at the same thing and then scale them together if the math works.

Posted Oct 15, 2007 9:52:01 PM | link

Thomas Malaby says:

That's not enough, if you don't look at what they do.

Posted Oct 15, 2007 10:59:34 PM | link

Richard Bartle says:

Dmitri>Richard, you ask if an unverified system is better than nothing, and I'd have to answer "Who knows?"

Well it depends on your definition of "verify". If someone has applied the theory and found that it works for them, they're not going to care whether there's been formal academic verification or not. (Yes, I do realise that people who believe in homeopathy, astrology and Zeus can make the same argument).

>If we also tied those measures to what they do or how they played, that'd be a pretty damn useful planning or testing tool.

No it wouldn't, not as it stands.

Suppose that we did find a link between what motivates players and how they play (which Nick's results don't even posit, but you did say "if" so I'll go along with that). What use is this information to us? We can presumably change the virtual world design to encourage or discourage people to play on the basis of which factor motivates them, but we don't know what will happen if we do that. There's no model here to explain it.

Why do the motivations exist? Why do they exist in the proportions they do. Merely knowing that the motivations do exist, and knowing what people who have those motivations typically do, doesn't help. Are they motivations inter-dependent or independent? Is it the case that creating a world aimed solely at socialisers (ignoring achievers and any immersion factor) would get more players than one aimed at all three? It could be that achievers dampen the number of socialisers, or that they increase them, or that they have no effect - which is it? The stats don't tell you. All they tell you is that there are three motivations people have for playing.

I can see some ways the data could be used for design purposes. For example, by analysing what actions players do, you could divine what motivates them, then dynamically give them more content of that nature. However, it could be that players change motivation rapidly, or they need a mixture of motivations, or it depends on how long they want to play. The data we have at the moment doesn't say anything about any of this.

If you want to use the data for non-design purposes, such as targeted advertising, OK, that's fine. My own theory is aimed at aiding designers, though.

I'm not saying it isn't possible to do build a theory in a data-driven way - of course it is. It's painstaking, incremental work, but it's not as if we're in any hurry. It'll take a lot more than tying player actions to motivations for it to be useful to designers, though.

>And, like I said before, it'd be more useful still if it added (i.e. added and then validated) the creative element Tim suggested.

And what if it didn't find them? Would these creative people then be illusory?

>I was under the impression that he'd checked on the killer and other factors and found they hadn't held up.

If you ask inveterate liars if they tell lies, how many are going to tell the truth? Some killers are indeed boastful, but given a set of survey questions they might take the opportunity to mess with their answers just for the pleasure of screwing you over. As a group, they could do that systematically for all we know.

I don't think this is necessarily an issue here (rather, I suspect the number of killers really is low, especially in the virtual worlds Nick tested). However, it does illustrate the kind of problems you get with these "validated" surveys. They're consistent, but not necessarily complete.

One solution, by the way, would be to ask a selection of non-killers to identify some killers (or, rather, people whose behaviour is commensurate with what the survey identifies as killer type), and then check out how many of them appeared in the survey as killers.

>The way to do this is to come up with a series of questions that line up with each of Richard's types.

This is what the so-called "Bartle Test" purports to do. Although this is somewhat flawed and not remotely rigorous, nevertheless the highest-rated player type it produces is Explorer - which doesn't even register in Nick's data, despite his adding more questions to his survey to try to give it a chance to show up. In the Puzzle Pirates version of the test, the highest ranked type is killer (!).

>If that works and the math checks out, you then have a tool for measuring how many people fall into these categories and who overlaps. And that's a working model that's very useful.

It's not what I'd call a working model, but coming from a computer science background perhaps I use the word more functionally than you do. As for its being very useful, well sadly, no, not for designers it isn't. We need to know why, not just what.

Richard

Posted Oct 16, 2007 4:15:01 AM | link

Timothy Burke says:

Another thing to consider here is the feedback loop that's involved in model-making or typology. E.g., when a particular model is impressionistically persuasive (or simply widely circulated) to the people who are modeled by it, that has an effect on how they conceptualize their actions and thus on their actions in turn. The Bartle typology is a great example of that. A lot of older players in particular have encountered it, found that it seems to describe some aspect of their experience in MMOs that was previously less clearly imagined, and then have practiced in MMOs from that point forward with an internal conceptual model of their own preferences.

A really hardcore positivist might say that this contaminates the phenomenon in question, I suppose. For me, that just seems quintessentially human: we make typologies and models the way we make metaphors, to gain a convincing but also compressed understanding of phenomena that are otherwise too big and complex for us to describe or conceptualized. Our metaphors and models recursively alter our lived experience, sometimes because they're a good fit to underlying realities, sometimes because they're an awkward fit. (Say, for example, the Freudian description of dreaming, which now seems to not be at all a good description of what's happening in dreaming.)

Yes, we need formal ways to push back on or correct models which have become influential but are badly fitted to some other underlying reality, and that's part of the job of social science. But even when we discover there's a bad fit, that doesn't mean the model is now worthless. The Freudian model of dreaming, for example, is still a creative and imaginative resource for many people, it still serves its purposes outside of being an empirical claim about what's "really happening". To cite another, similar case, it's increasingly clear that the standard model of human motivation in economics (maximization of utility) is just plain wrong as a description of how people really think and act in the world--but it's foundationally necessary for a tremendous amount of useful microeconomical knowledge.

Posted Oct 16, 2007 8:36:32 AM | link

Anders Hall says:

This discussion of models is quite similar to one I recently had in Uni before my (master) graduation. My science teacher (during 20 weeks of pure scientific theory) reflected on models of human behaviour in relation to Interaction Design (which is my trade). He argues that models of human behaviour often are problematic because they have a hard time taking into account how we as human change over time. When the model is close to finished reality has changed. For instance, a complete model of human behaviour is therefore not realistic since we change over time and also react in relation to the model (if we know it exist). ..philosophical thinking of course. In Interaction Design we (to some degree) try to catch those, hard to measure, aspects in real time and apply them in design (e.g. with player studies/user studies).

I mention this because I think the value of your discussion might be (as Mr Bartle indirectly state): Mr Yee's model does not deal with the smaller groups which are harder to test and discover, it is a descriptive model that always will lag behind reality. Mr Bartel's model is, in my opinion, more ambitious, trying to be "up to date" and very usable. Has any model, as of today, achieved such a ambitious goal?

The point I am trying to make is that the creative aspect have always been there in virtual worlds though both models mentioned have failed to address this "factor" to any greater degree since it was not that obvious to spot.

Virtual worlds are persistent and, in my opinion, evolve faster than the real world. Lets say - compared to my neighbourhood :-) . At the same time the people playing, or perhaps designing, change and so does the real world (as Mr Grøstad mentioned). Even more aspects to take into account.

The artists, technical staff, designers and gurus which design and build virtual worlds have notices this creative "factor" and brought it out in the open - before both models describe it! Both models will, in the same way as the Linnean taxonomy, change over time. But how can you catch new phenomenons quicker than today? I think Mr Bartle will need more staff :-) and Mr Yee might continue with a slower pace (and more in depth testing).


PS: I do not see my self as an artist (some Interaction Designers do). Though that last part of creativity, following user/background studies, is hard to explain in scientific terms. I think that categorising game designers as artists is fair - otherwise we will need to stop using the phrase artists completely (which might be mandated though boring for us all).

Posted Oct 16, 2007 9:07:44 AM | link

greglas says:

Tim is absolutely right that this is a recursive system where designer conceptions of players inform game designs which enable the players to conform to the designer conceptions of their behaviors. That is just one of the reasons why I think of the Bartle types as a designer's rule of thumb, more important as a way to lead designers to appreciate that there are a diversity of play styles than as any kind of empirical truth about play practices.

Years ago, I had a high school art teacher, Claude Falcone, who had us critique each other's paintings, etc., by applying a list of 7-12 (it varied) criteria and rating how well the work did things such as 1) maintain a balanced composition, 2) avoid "wasting" space, 3) provide a compelling center of interest, 4) utilize sufficient contrast, etc...

Of course, what we quickly understood was that while successful artwork would generally meet the criteria and it was extremely helpful to be aware of the criteria as an artist, *all* the rules could be successfully broken and meeting the criteria was only the smallest part of what made a piece of art visually compelling.

The key utility of the Bartle types, I think, is how they enable designers to design. For a person focusing on a different issue, such as evaluating the degree of social interdependence in an online community, they might not be very important.

Posted Oct 16, 2007 10:05:13 AM | link

Dmitri Williams says:

Thomas, we can and do look at what they do. That's the point of the server-side behavioral data that can be combined with attitude measures, e.g. measuring how often someone ganks someone X levels below them might be a pretty good indicator of a "killer," e.g. And, this approach checks whether they are accurate in their self-assessments (answers to scale questions like Nick's) and then whether and how they change over time. You can have people answer the same questions at later points in time to see change.

And not everyone lies on surveys. Not even the assholes. We now have decades of built up knowledge about how and why and when people answer with accuracy, and how to craft questions and batteries that do in fact work. That's the verification ("Validity" in my jargon) I've been talking about.

Answering the "why" they do what they do is a job for both survey work and ethnographic work done together.

And of course this is all terribly complex and a moving target. It's people. So what? Just because it's hard doesn't mean it isn't possible.

Does no one think that having proportions for what people do, how much they enjoy it, why they do it, and why these things change is useful? Sure, the Bartle types allow design, but doesn't the design change if the true baseline % of killers is 1% vs. 25%? Or if we know that all players have some degree of it? I must sound crazy on this thread, but I like data and I like knowing these facts. I think some of you fear that these measures are part of some Borg-like soul-crushing corporate ethos that is the enemy of art, creativity and love. They're tools, which can be used for good or ill like any other, like say Freudian analysis. They're also used with varying degrees of skill, like any other approach.

Posted Oct 16, 2007 1:35:44 PM | link

greglas says:

Well, personally, I don't know about the rest of you all, but I *don't* think Dmitri's measures are part of some Borg-like soul-crushing corporate ethos that is the enemy of art, creativity and love. (Although if he were doing this for EA, that might be true.)

Just so I'm not misinterpreted here -- while I like the Bartle types, I'm incredibly interested in knowing what kinds of data Dmitri finds useful and what kind of models he creates.

I don't see anything problematic in both defending the usefulness and constructive insight of the Bartle types and at the same time allowing Dmitri to say that, given his preferred methodology, he finds them problematic.

Posted Oct 16, 2007 2:01:18 PM | link

Thomas Malaby says:

Dmitri, we agree on 99% of what has been said here -- I think that Tim and I (and perhaps Greg), are making small objections about the language precisely so as to avoid reproducing the ways that certain valid research approaches have been marginalized in the academy in the past.

I've certainly never said that I don't think such typologies are useful -- on the contrary, as every comment I've made shows. And of course you look at what people do -- I never expected you wouldn't -- it's just that I'm quite a bit wary of researchers taking representations on surveys, in interviews, etc., at face value. So, while you quite rightly look at practice as well as representation, you then also suggest (and maybe I'm misreading) that surveys in themselves contain enough checks and balances to make the critical concerns something like a non-issue.

So, to take the representations people make on surveys as the example at hand. To be critical of them doesn't mean that we need to suspect that the informants are *lying*. In a way, this way of conceiving what the danger is may be the problem, because the real challenge is that they may *not* be lying, and yet may *still* give answers that diverge from their practice. And as regards *validity* with respect to the representations that people give in interviews and surveys, it's worth remembering that every one of these well-established empirical techniques that gather representations (like surveys, structured interviews, semi-structured interviews, open-ended chatting) has similar well-developed techniques for checking answers against each other, against others' responses, and (in some ways, most importantly) against the often changing status of the researcher him or herself vis'a-vis the informant. We're all part of the big tent of professionalized social research, after all -- but sometimes I get the sense that certain methodologies are assumed to be less thoroughly professionalized and refined. My apologies if that's not the case, of course.

Posted Oct 16, 2007 3:06:47 PM | link

Richard Bartle says:

Timothy>when a particular model is impressionistically persuasive (or simply widely circulated) to the people who are modeled by it, that has an effect on how they conceptualize their actions and thus on their actions in turn.

Yes, this is true, although I doubt that more than a tiny fraction of players will have read my paper and adapted their behaviour to fit the slot they perceive as most prestigious or whatever. I know that none of the players in any of the WoW guilds I've been in have known who I was when I told them my name. I'm not actually famous.

>A really hardcore positivist might say that this contaminates the phenomenon in question

The same person would probably decry Nick's early work, then, on the grounds that if you survey an achiever-based game, you're going to find a lot of achievers in it and smaller numbers of the other three types. Taken to the extreme, that would mean it was inextensible beyond EverQuest.

Richard

Posted Oct 17, 2007 3:41:55 AM | link

Richard Bartle says:

Greg>Tim is absolutely right that this is a recursive system where designer conceptions of players inform game designs which enable the players to conform to the designer conceptions of their behaviors

This isn't what Tim was saying, but it's right anyway. If the designers design for four player types, then it shouldn't come as a surprise when they get those four player types. Of course, if the model is complete and correct then this is a good thing; if it's missing people or mischaracterising them, then it's less useful.

Remember, though, that when I wrote that paper we were seeing virtual worlds created for only one person - the designer. Vast swathes of potential players were being put off because they didn't share the designer's idea of what fun was. We were getting into the feedback loop you describe anyway: design for an achiever, you attract achievers, and when some of those achievers themselves become designers then they also design for achievers, only even more so. At least my paper raised the possibility in designers' minds that there were other types of player at all; if that's its only enduring legacy, I'd still be pleased.

Richard

Posted Oct 17, 2007 3:52:23 AM | link

C-Park says:

Richard points out designers build VWs they'd like to play in. Players try to make the designer's vision their own. Then Nick studies those worlds and sees echos of the designer's subconscious filtered through player perceptions. But many of us speak about VWs (perhaps not deliberately) in grander terms, as though they are somehow reifications of human conciousness that escape the bounds of tawdry physicality. Could it be that by it's very nature that's too big a bite for us to take? If that's so, should we put more up-front effort into limiting the scope of our work? If so, how can we do that without straitjacketing ourselves?

Demitri described the measurability of player behaviour when he said "That's the point of the server-side behavioral data that can be combined with attitude measures, e.g. measuring how often someone ganks someone X levels below them might be a pretty good indicator of a "killer," e.g. And, this approach checks whether they are accurate in their self-assessments ..."

If, broadly speaking, we can measure three things: what actions players take, what things they acquire, and what they report when asked -- then we are constrained to characterize players in ways that those measurements can document. Why? Because we are trying to create an environment or effect a change and if our work is unmeasured we can never know if we succeed. No MMOG left behind!

Like that famous government program, however, we can run afoul of the definitions driving the content. That is most likely to happen when we over-constrain our goals.

Richard speaks to that point when he says "If the designers design for four player types, then it shouldn't come as a surprise when they get those four player types. Of course, if the model is complete and correct then this is a good thing; if it's missing people or mischaracterizing them, then it's less useful."

If, as Richard suggests, we define our design objectives (perhaps iteratively) carefully then we are free to measure against our own, rather than an externally imposed set of metrics. We don't need an over-arching scheme to be successful. Saying that, brings us back to our original question.

Where should the edges of our efforts lie? The passion around discussion of Richard's and Nick's work surfaces how much we all want to understand MMOG game design's defining elements. Can we understand them without getting caught up in understanding why people do anything they do?

Posted Oct 17, 2007 10:31:35 AM | link

Dmitri Williams says:

Paraphrasing something I said to be akin to NCLB makes me want to /wrists.

There's a solution to the issue that doesn't constrain the model to the proclivities or imaginations of the measurer. You start by asking people why they play, and you take open-ended answers. You ask a lot of them, and you collect a lot of answers. Then you add anything else you can think of. You then go over them to knock out similarities and find the minutia. Then you make that large list of 100 or so items into an exploratory study on a random sample of players and see how they factor. In this method, the players drive the measures, not the measurer. I don't know if that's how Nick did his, but that's the solution we learn in survey methods, and it's how I'd do it.

Posted Oct 17, 2007 11:25:58 AM | link

Thomas Malaby says:

It's also (if differently achieved) the heart of the ethnographic enterprise, so we very much agree. Exploratory methods FTW.

Posted Oct 17, 2007 11:30:24 AM | link

C-Park says:

Dmitri, my sincerest apologies for writing something that you could interpret the way you did. I had no intention of suggesting that your methodology (which I believe I understand and agree with) necessarily led to a NCLB attitude toward the insights the methodology provides. To Richard's point about studying something that was created within preexisting constraints on content and the player's it attracts, however, I do believe there can be a self-fulfilling prophecy. To the extent that any ethnographic work studies a particular population it can only expose the characteristics of that population. Others can take those results and use them improperly. As a counter to that, if enough different populations are explored, as I expect you try to do, then commonalities and differences between them can be surfaced.

It had been my hope to suggest the difference between those common characteristics and individual game specific characteristics could lead to a better understanding of the scope of VW / MMOG endeavors provided the range of populations explored were not artificailly constrained by the traditional procleveties of current game designers. Additionally, I wanted to point out that without some agreement on the bounds of what is or is not our domain of investigation the enterprise could get out of hand, especially for those of us whose motives are tied more closely to the day-today of game design and less to exploring deeper aspects of players' involvement in games and VWs.

Once again. My apologies.

Posted Oct 17, 2007 12:26:36 PM | link

Dmitri Williams says:

No offense taken at all. I just happen to be married to a schoolteacher and I hate that legislation.

One of the advantages of survey work is that if you get a random sample, you don't measure a subgroup. You measure the whole population. So if your instrument is good (always an "if"), you get a census of whatever you're after. It's the breadth and generalizability that ethnographic work can't manage, which makes it an ideal complement to the depth that ethnographic work often does add.

Posted Oct 17, 2007 2:13:58 PM | link

dmyers says:

Hi Dmitri,

We had a conversation about this at some distant conference, but I don’t think I made any point particularly clearly. You have made yours clearly, I think. It is, I believe, a fairly common point: that (generically speaking) qualitative research is best used as a preliminary step to (generically speaking) quantitative research. There are, however, dissenters dissenting (best, I believe) along two similar lines.

These contrary arguments are that qualitative research can’t really be usefully extended by using quantitative methods because 1) people are unable to articulate, through either behavior or language, their motivations (or whatever it is you want to find out about them), and/or 2) their (or the researcher’s) articulations are only valid and useful within the context within which they can be articulated. These arguments can be related, but they don’t have to be.

Argument one is based on the notion that there are forms of knowledge and understanding that require human experience and that these forms cannot be usefully shared through sign/symbol systems, whether those systems consist of words or numbers. This is sort of a phenomenological perspective, though those who would call themselves phenomenologists (or maybe mystics) seem non-uniform regarding it.

Argument two is based on the notion that we can indeed learn something from, say, the scientific method, but that those things were learn are, to our simplistic minds, gobblety-gook. An analogy might be uncovering some sort of correlation values that we then try to varimax rotate into “factors.” This attempt to “understand” data then renders them progressively meaningless. A Cthulhu thing.

Argument one implies we are isolated individuals. Argument two implies we are an isolated species. In both cases, however, we are isolated by our dependence on symbolic forms and the limits of those forms. Those forms could indeed include bodily mechanics and senses.

Statistical data gathered from large groups tends to show some characteristics that could support these arguments. In the first case, we as individuals tend to consider our position in a large group, or in the world, as under-defined by statistical representation. In the second case, we as scientists tend to under-understand and, simultaneously, over-value our interpretations of empirical measurements, e. g. the medieval astrologer.

Just saying.

Posted Oct 17, 2007 3:30:44 PM | link

Dmitri Williams says:

DIGRA, Vancouver, and I remember it well :)

Notwithstanding the two arguments here (which to my addled brain seem a bit like crossing the streams in Ghostbusters), I like to see qual & quant go back and forth with no one method being a particular "final say." After all, survey work can unearth a lot of "why" questions that might then be best resolved by ethnography or critical analysis. Sometimes a survey is too blunt an instrument.

As an example, I have some really nifty survey data on role players, but it prompts a bunch of "why" and "how" questions that I don't think are best answered by a follow-up survey. And the target population is fairly small, so a more hands-on and in-depth ethnographic approach seems like the next step. And that's what I've recently put in motion.

-D

PS I nearly always use oblimin rotations since varimax can assume orthogonality when humans are rarely so ;)

Posted Oct 17, 2007 4:50:04 PM | link

Douglas Finnigan says:

I'm teaching an introductory subject on 3D game development at a Singapore polytechnic, and covered the idea of typing, including Richard Bartle's model. The students loved it. They said it really made them look at (their own) gaming in a new way. Immediately, their own work went from vague to informed, analysing games and game scenarios in terms of how much they supported the different types. I'm no expert, but whether or not the model "works" or is testable, I can see the effect it has in an educational context - it gets the students fired up about abstractions! This was a very pleasant surprise to me. I doubt if any model of human behaviour can be exhaustive or ever completely validated - but if it gets us thinking in a logical and structured way about what is previously considered to be illogical and unstructured, then it's already worth the paper it's written on.

Posted Nov 18, 2007 7:45:54 AM | link

Adam Hyland says:

^^^^
ZOMG spam.

Posted May 12, 2008 3:44:19 PM | link