Sunday, November 10, 2013

None Of The Above : The Bartle Archetypes

Jeromai had some very interesting things to say about the current World vs World Season, tying neatly into the ongoing debate on Achievements. While I was reading it occurred to me that one of the foundations of the whole discourse might be developing a slight wobble, namely The Bartle Test.

We all know the provenance of this, I'm sure. Anyone who doesn't can get up to speed in a minute or two at Wikipedia. We all quote the Archetypes and our own four-letter formula and score whenever it seems relevant. What I don't know and have never really even thought about before today is whether it still merits the attention we give it almost twenty years on.

Richard Bartle posited the four archetypes, Explorer, Achiever, Socializer, Killer back in 1996 and we've all just gone along with them ever since. Do we really think those are the only four motivations driving players? Are they sufficient to explain all player behaviors in all MMOs over the last two decades? Do they accurately, sufficiently and exhaustively describe the entire and full range of psychological factors that drive every individual to play and to keep playing these games?


Doctor Bartle didn't pluck the archetypes out of the air. As he says in the preamble to the very readable, chatty paper that started all this, they were condensed from the comments of several hundred posts on a bulletin board belonging to a MUD, a lengthy conversation to which some thirty people in total contributed, the bulk of the comments being made by half that number.

The four archetypes having been distilled from this boiling pot of argument, some time later two entirely different individuals developed the Test we all know and which we take and retake just in case we might have had some kind of personality change since the last time we tried it. The test itself reminds me of nothing so much as those "Is He Really Right For You?" quizzes so popular in teenage magazines of the 1970s and 80s, in which you would frequently end up checking a box against something you would never in a million years say or do simply because the other options seemed even less plausible.


Whether there is any scientific validity in this methodology per se I have no idea. To what degree this kind of psychometric testing, if that's indeed the category under which it falls, can be considered to be part of the Scientific Method, I couldn't say. If there's any kind of recognized peer review process for this sort of thing (if there even is a "thing" for this to be a sort of) I wouldn't know.

Either which way, it seems odd that one man's summary of fifteen people having a heated discussion on a MUD bulletin board almost twenty years ago (with another fifteen kibbitzing and chiming in from the sidelines) should have come to form the basis of design decisions for an entire genre of entertainment that didn't even exist at the time the discussion took place.

I never played MUDs, having graduated University in 1981 and having anyway read English, meaning I never even saw a computer let alone used one. It was a little after Richard Bartle wrote his first paper that I got my first home dial-up and by the time I was ready to consider using it to play games online Everquest had been out for over six months.


Consequently I can't say from personal experience whether the Bartle Archetypes accurately represent the full range of activities available to MUD players. I would contend, though, that even if they did they fall some way short of describing the options available to players of modern MMOs.

It's true that one can almost always collate any number of human activities under larger and larger catch-alls so no doubt an argument can be made that KASE covers all bases, but I wonder if such reductio ad absurdum methodology really gets us all that far? When I spend four hours on a Sunday morning sorting my banks, something I've been doing for many years, which archetype is that, precisely? When I do my dailies with no intention whatsoever of using any or the rewards but just because they make for a pleasant, soothing, familiar routine that relaxes me when I get home from work, where does that fit? When I swing the camera about while I run up and down the ramps in Black Citadel as a Charr for the umpteenth time, imagining I'm really a giant cat, what am I exploring, achieving, or killing?

I know I can't be socializing because there's no-one there but me and isn't that another thing that's changed out of recognition since the MUD days? So much of one's time in a modern MMO can be spent entirely alone, and amusingly so. If I spend a week moving the furniture around in a virtual house that no-one but I will ever see, or working nostalgically through quests I've done a dozen times before, this time on a character that I know will soon be sent back into the limbo of character select, half-forgotten, rarely if ever to be called upon again, which archetype am I fulfilling? In spending hour after hour concentrating on visual images while playing alone, am I even playing in the same ontological space as that inhabited by the text-adventuring, community-oriented participants of that seminal bulletin board discussion all those years ago?


Next year we expect to see the launch of a multi-million dollar, triple-A MMORPG overtly and proudly based on Archetypes : WildStar. Carbine stuck with four but kept just one of the names; Explorer. The other three they changed to Soldier, Settler and Scientist. Whether it's a nomenclature that represents the behavior of the current wave of MMO players more accurately than Bartle's established quartet we will find out in due course. I rather doubt it.

The more I think about the whole thing, the less I feel represented by any of the terms in play, old or new. Where's Potterer or Observer? What about Collector or Organiser? Photographer or Sightseer? If we can, as is entirely possible in many MMOs nowadays, play for hours, days, even weeks without exploring anywhere, achieving anything, killing anyone or communicating with another human being, and yet still have a really fun time doing it, mightn't it be time to lay the old archetypes to rest?

Or at least come up with some new ones?


  1. ah, the Bartle descriptors: I simultaneously love them and grate against them. Love them, because as high level population descriptors - which they are - they have a high degree of descriptive utility. Based upon my reading of your blog and those comments I've seen on other blogs I read, I'd be quite comfortable as saying you are generally within the Explorer population.

    Grate against them, because in spite of the fact they're good population descriptors, people insist on using them as individual motivation descrpitors - which Bartle himself has repeatedly tried to correct, but without success. Nick Yee's entire research career started (what feels like yesterday, but is now over a decade ago) because he wanted to create actually useful *individual motivation factors* - I'd recommend doing some of his motivation surveys, if you're dissatisfied with the Bartle descriptors :) .

    It doesn't matter that I find the Bartle quizzes amusing (and give them to my students, as they're fun ways to show them how complex profiles can be generated using just either-or questions), they're still basically wrong: attempts to use population profiles to explain individual motivations; which are, frankly wrong. Which has, inter alia, made me suspicious from the beginning about Wildstar: if it's based it's classes on a misunderstanding of the Bartle model from the beginning, then things can only end badly. Designing for the four population types can be a sueful shorthand; designing your classes as if each individual is always and only part of one population type: that's never going to end well. It is completely normal for any one individual to a: never identify with a given population descriptor; and b: be identifiably part of one population much of of the time. Thus: I'd be quite confident situating you as typically within the Explorer population (but not exclusively: that'd be the point) without you *ever* identifying as an Explorer. Bartle *has* got a model where he adds an extra dimension to his population descriptors to try and generate individual character types; but for whatever reason it hasn't resonated the way his basic model has.

    I'll note that game designers get this as wrong as players do: Wildstar's 'population = individual character type' impending train wreck is hardly alone; you've documented repeatedly GW2's mistaking of Explorers as Achievers (eg, via the use of achievements - or however GW2 names them - to motivate players to explore the deepest ends of the map; missing the point that the Explorer population is internally motivated to explore and typically resents external motivations) (although arguably they decided that Achiever-based populations a: require external motivation, that's the whole point; and b: are a much larger population than Explorers; thus they catered to Achievers deliberately, even though ot de-motivated Explorer-types) (and yes: that's too many nested brackers *and* it's extremely easy to slide from talking about population models to imputing individual motivations, as I just did).

    At any rate, as a hopefully clearer summary: Bartle's model is a high level population model; not a description of individual motivations. Expecting Bartle's model to cover the full range of individual motivations is thus flawed: whether we do it as players, or whether MMO designers foolish enough to build their character classes around the model do it.

  2. Thanks for the gloss! Very helpful indeed. I did suspect it was something along those lines.. No-one could reasonably deny the usefulness of the archetypes as general indicators of direction of travel for segments of the MMO population; it's when they come into play as overt, conscious motivations that doubt creeps in.

    I contributed to some of Nick Yee's surveys, including the first one that came out not all that long after I started playing Everquest. You'd like to think what with over a decade of academic research plus all that metric-gathering that goes on automatically in modern MMOs the developers would be in a position to predict player behavior with some degree of accuracy by now, but I see precious little sign of it.

    Probably just as well, really. I scarcely know what I want out of MMOs from one day to the next so why should I expect anyone else to know?

  3. Another thing to remember is that Bartle's model also measures things on a scale, but out of convenience when we simplify and make abstractions, many people treat them like separate dichotomies - which they actually aren't.

    In my opinion, we all have a little of A, E, S and K in us, just different amounts, and for many, we evidence distinct preferences in our MMOs, so when addressing player categories as a population in MMOs, one tends to just lump all Achievers as one group, all Killers as another, etc.

    There have been many attempts to develop Bartle's model further. Some use a combination of the declared primary and secondary preference to attempt to distinguish between say, an Explorer-Socializer and an Explorer-Killer. I think Bartle himself added intrinsic-extrinsic motivation as another dimension to the axises.

    Another caveat is that some players have been known to move between the various axises, depending on situation and time. There's a theory that tries to predict where players fall within based on their experience with the game, but I personally think that's trying to generalize a bit too far to the point of not being useful.

    The other thing to recall is that the model was primarily used in the context of MUDs (and a distinctive type of DIKU at that), and then extrapolated to MMOs, where it mostly only fits insofar as MMOs still resemble MUDs. If I'm playing an FPS, I'm going to be evidencing very Killer-type of behaviors since those are the rules of the game that I've accepted. For their purposes, it may be more useful to segment player categories using another model, such as offense-preferring, defense-preferring, attack/support or whatever.

    Nick Yee's motivations of play catches more of the categories that Bartle's model missed for MMOs.

    For some of those activities you describe, I'd hazard a guess they fall under Immersion, possibly Escapism because they relax you. And others under Roleplaying because you're imagining yourself a giant cat in a fantasy world.

    There's another meta-synthesis I found that tries to synthesize further previous papers on player types from Bartle on, which gives more useful summary/coverage from another angle:

    1. haha.. I was actually replying something like this in your post. It does seem to be more of a combination of factors that go into it, there might be pure ones there but in some way having a secondary changes the core approach of the person. Those white hat pvp folks would be more the social killer types.

      Also that linked bartle test is stoooooopid, so many badly thought out questions that neither give a good indication of the type or are just confused questions. A magic staff to gain experience, what... you would think the achiever type would want to earn it.

    2. For all the academic research into online gaming that's gone on over the last decade and a half, it seems surprising that the Bartle paper and the subsequent Test remain far and away the most widely cited and used by players and developers. I used to follow the Terra Nova forum where this kind of thing is discussed on an academic level routinely so I was aware of the depth of research behind the scenes, but rarely does any of that surface in the mainstream gaming media.

      I remain generally sceptical of most of the social sciences and my scepticism only increases the further away from the mainstream you get. I suspect that ten, fifteen or twenty years ago research into the behavior and psychology of online gamers was attracting neither the heaviest funding nor the most ambitious researchers. That may have changed as both "online" and "gaming" have become mainstream but if so whatever new research may be going on doesn't seem to be making much headway in the popular gaming media, or in the offices of the people who make the games.

    3. and coming back here several days later, a few more thoughts occur to me:

      1. There is a lot of work done on player motivations, but most of it is not in the public domain: I conducted a very large (and, I believe) thorough research project five or six years ago now, into player motivations - or more specifically, into individual drivers and barriers to computer game usage. It was for a large project to design a new console that never went anywhere, unfortunately - and the research I did was swallowed up by the companies funding it, and never got published.

      I'll note that it was statistically solid, and gave a fair bit of support to the basic premises of the Bartle model (understanding, of course, that individual motivations are only tenuously tied, at best, to population models) - sufficiently so, that _if_ you were a game designer, and you conducted similar research, you'd probably find your immediate superiors questioning why you spent budget when the Bartle model was already known.

      2. Due to a variety of factors, there has been a real turn to causal analysis of player activity - A/B testing and the like - over the last 5-6 years. This sort of research generates easily actionable conclusions and predictions, can be conducted in real time on the player population, and is readily explainable to immediate superiors. It ends up diluting diversity, appealing to mass markets and commodifying gaming and play, but oh well. Easy metrics are very seductive, and because of that, descriptive work such as Bartle's or Yee's has fallen out of favour amongst gaming companies: A/B testing and the like is so easy to conduct and to explain, that larger questions of motivations have taken a back seat.

      3. Motivation research is rather like market research: many people sing it's praises but far fewer actually do it. I've got the impression over the years that many gaming companies believe that, because they themselves are made up of gamers, they know already what (other) gamers will want, and thus don't do any research. It's notable in this context that Wildstar namedrops Bartle: I do think they're doing it wrong; but nevertheless it could be evidence of a seachange; of the need for gaming companies to have some sort of 'scientistic' justification of their business model. There are quite a few MMOs that I can think of that could have benefitted from even a mistaken understanding of gaming motivation models (whether population or individual).

      I don't think, however, that we'll ever get the 'heavy hitters' doing research into the field: the prevalence (and ease) of A/B testing means that the funding isn't there; while the general decline of the MMO as a genre (this is not meant as a lament; I just happen to think that MMOs are done - especially fantasy MMOs - in large part because we misunderstood why they were briefly popular in the first place) means that the pressing need for the research isn't there. There's lots of research being done into social gamers, mobile gamers, tablet gamers, etc: but it's almost all in-house, it's almost all causal, and it's almost all linked to ever-more-efficient monetisation of users...

  4. I really should take that bartle test. Also that's actually a pretty good Wonder Woman costume. :P

  5. It's all very nebulous. I came to WoW an achiever, achieved exactly nothing and found I liked exploring.

    I came to FFXIV an explorer and am just socialising all the time. I have no idea what I am anymore. I need therapy. I need to be straightened out.

    The War On Failure slayed the achiever. The War On Redundancy murdered the explorer.
    I'm pretty sure socialising is immune to these forces as long as we still have chat windows.

    Oh God, they wouldn't take those away from us would they!

  6. It's very appropriate to be questioning the Bartle types.

    In my opinion the four type scheme is so limited to be almost worthless. I've never felt that I fit in to that scheme.

    The eight type scheme is somewhat better, but even that seems lacking. The Bartle types seem too oriented towards theme parks as opposed to sandboxes. It's unclear in which categories artists would be found.

    I think the appeal of Minecraft would indicate that Bartle's categories underrepresented the creative types.

  7. Horrifically late commenting here, but just wanted to echo what seanas said above: a lot of the confusion is based on misapplying the Bartle motivations and simply misunderstanding them. For example, Killers are PvPers, sure, but Killers can also be the strict leader type that keeps people in order. Killers like to act upon other players, and that takes various forms. Also, players rarely have a single motivation, and player motivation can change a lot over the course of their time in a game.

    The best way to use the Bartle types as a developer is to figure out what someone with a strong motivation in an area will do. If a Killer comes into your game, what will do they do? If you don't provide any sanctioned gameplay, they'll likely become griefers which cause customer service headaches.

    I haven't really embraced Yee's work because I think the data he accumulated was for a very specific game at a very specific part of its life cycle. In particular, I think his lack of finding an explorer motivation was because EQ didn't particularly cater to Explorers, especially after all the initial areas were mapped out. The game catered heavily to Achievers, which meant that changing the game and mechanics was something best done in careful measure. The rise of the game database sites (created by Achievers to further their achievement) also reduced the role of explorers.

    And, yes, some type of play is going to fall outside the motivations. Running around pretending you're a giant cat? That doesn't fulfill a game purpose, so it falls outside Bartle's classifications. Tweak that activity a bit and it can; if you were pretending you're a cat to role-play with others, then it become Socialization. Also, the question about inventory sorting is interesting, particularly since I'm the rare type that actually likes inventory sorting and limitations. :)

    1. So late you fell foul of the "moderate after 14 days" function!

      I agree with the observation on Nick Yee's research (other than the idea that Everquest didn't cater for explorers - I would say it was and still is an explorers' paradise) but isn't that exactly the problem with the original Bartle archetypes? They are drawn from the behaviors available to the players of that particular MUD. MMOs would seem to me to include a lot more options nowadays that don't comfortably fall into any of those categories.

      The giant cat thing is a specific example of that. The MUD Bartle was drawing from was pure text. The categories do not and could not include any purely visual activity. Running characters around, swinging the camera about and watching them move is something I do in many MMOs and something which forms a significant part of my motivation to play one game over another. On of my earliest posts on GW2 was an exultant account of running my Charr up and down the ramp in Black Citadel just to watch how he moved. I even took a video. Taking screenshots is another visual activity, and taking screenshots has nothing, necessarily, to do with exploring. It's an activity that relates very closely to taking photographs.

      Similarly sound and music. Mrs Bhagpuss frequently goes to specific places in MMOs and does activities there because of the in-game music at those locations. I don't believe MUDs at that time would have had full orchestral soundtracks.

      I think Bartle's archetypes are all still present and relevant in modern MMOs but I don't think they cover anything like as much ground as they would have when he came up with them.


Wider Two Column Modification courtesy of The Blogger Guide