Having an empirical framework of what motivates people to play MMORPGs provides the foundation to differentiate players from one another, as well as to explore how these differences shape in-game behaviors, how these groups of players interact with each other, and how the design of an environment impacts its appeal to different players.
Richard Bartle postulated a framework of 4 types, and I’ve tested the model empirically by running a factor analysis on a battery of “motivation items”, but I’ve never been happy with those results. I reanalyzed the data recently after understanding factor analysis techniques more, and realized I should have been using oblique rotations instead of orthogonal rotations. But then I also fiddled with using a principal components analysis versus a factor analysis (MLA).
The PCA/oblique analysis revealed 6 elegant components that were very interpretable:
1) Relationship: Make Friends / Offer and Give Support
2) Grief: Scam / Taunt / Annoy / Dominate
3) Immersion: Story-Telling / Role-Play
4) Escapism: Escape / Vent / Forget RL Problems
5) Achievement: Achieve / Accumulate Power
6) Analyze: Rules / Mechanics / Mapping
The problem is that the “Analyze” component doesn’t show up in the EFA/oblique (the other 5 do + a Leadership factor) suggesting that the “Analyze” component is explained by the remaining 5 factors, and doesn’t contribute to explaining the correlations between the “motivation items” even though one could create an “Analyze” component and measure it reliably. So there is and isn't an "Explorer" type depending on how you want to look at it.
One interesting finding was that the “Relationship” component explains a difference of 10 hours (bottom of page) between players who scored in the top versus the bottom quintile of the component. The “Escapism” factor came next, explaining 7 hours. So, the “drive to achieve” doesn’t appear to be a primary differentiator between low and heavy use players. Now, they may be hanging around but they’re not the “power” gamers, just that they spend a lot of time in the world. The other elegant thing about this framework is that it applies well even to more social MMO environments like There.com.
Thoughts? Comments? Other potential motivators to test?
Urk. This suggests that reducing gameplay session length by tackling slow gameplay and timesinks will not be that successful in lowering average hours played per week.
Nick, as long as you are throwing around free research...
- what factors had the LOWEST time online?
- what factors were most common?
It's been a piece of commonly held wisdom that one of the keys to opening up the genre to a wider audience was to lower average session lengths and thus hours played per week, on the grounds that the wider markets do not have that much time to spend.
When UO put in power hour and the 8x8 macroing code, its average hours played per week fell by around 40%, without a loss on active users. That was generally seen as proof of the idea. But your new data makes me wonder.
Posted by: Raph Koster | Nov 29, 2003 at 01:34
Where does problem/puzzle solving or whatever it is that makes Solitaire and Minesweeper fun and replayable for many many people fit in? I'd say Analyze, but it doesn't seem to fit with what you were measuring for that category.
Also I'm guessing creation/building gets shoe-horned into Achievement somehow?
I'd be curious to see if you can extend the analysis to single-player games to see how the motivations may differ. Some categories I'd expect to be enhanced by sharing the experience with others (Relationship, Grief, Achievement and wherever creation/building goes) while perhaps the other categories would get less of a boost?
Posted by: Bryan Allman | Nov 29, 2003 at 03:03
Raph, I think the idea still holds, if you rephrase it "one of the keys to opening up the genre to a wider audience is to lower the *competitive* session length." This more accurately addresses the casual player's frustration with the power-gamers (being able to play so much and get ahead). If the marginal utility drops significantly for hours invested beyond the point of "casual gaming", this will always reduce the amount of time players feel they need to put in to compete. This is the hurdle for the casual gamer, as I see it.
Since this mainly applies to the achievement factor, I don't think that high average hours/week due to social aspects means dissuasion of casual members. Socializers just aren't measuring themselves against the other players in the way achievers are.
Posted by: Tek | Nov 29, 2003 at 06:47
True, Tek, but it'll still serve as an adoption barrier if only because of word of mouth about the typical hours played. We get a lot of people as it is saying "I'll never get into that sort of game, it'll eat my life."
Posted by: Raph Koster | Nov 29, 2003 at 11:49
Difference in hours of usage per week between top and bottom quintile of 6 components (average hours per week for top quintile):
Relationship: 9.8 (29.5)
Escapism: 6.9 (27.8)
Achievement: 4.3 (26.5)
Analyze: 3.5 (25.5)
Grief: 2.2 (25.5)
Immersion: 1.2 (24.4)
It's not possible to directly compare the scores of one component against the other because it's not clear whether 1 unit of Relationship is comparable to 1 unit of Grief.
The other thing re: Raph's "grind" point is that all these people are on the treadmill and the "Achievement" people are only your power gamers, so lowering the grind index will still have the effect of making it more accessible to the masses. The point is that you want to reduce the grind index for the people who don't score high on "Achievement".
Posted by: Nick Yee | Nov 29, 2003 at 15:12
Bryan, that would fit under "exploration," which encompasses not only the exploration of the physical-virtual world, but also an exploration of the physics that make up the world. "What happens if I go over this hill? I see this lake! Pretty! What happens if I cast a spell wearing chainmail versus plate mail? Lower spellcasting time! Interesting!"
I'm curious if there's a perceived difference between storytelling and interactive roleplay. I was always under the assumption that "roleplay" involved in-character reactions to the world and its inhabitants, but these days, "roleplay" is generally assumed to mean fabricating a private story with your friends, and playing it out, within the game's mechanics but independant of the game world.
Posted by: Slyfeind | Nov 29, 2003 at 16:02
In order to achieve, one has to optimize XP rate, which means you usually have to do some work with charts, tables and such. I see such "exploring" as a natural part of achiever behavior.
I think Yee's study is lacking questions which force player to choose between achieving and "exploring". If my idea of explorer finds faberge egg, which she could transmute to stupendous amount of gold, she would feel horrible urge to smash the thing to see if there's anything inside. Also, she would find it fun to *spend precious exping hours* to see how would hunting fellow of 30 druids do, if everyone just summoned a pet kitten and let the feline army do all the fighting.
Although there's a chance to find something even more expensive inside the egg, it's a very slight and costly chance. I would call player explorer if she's knowingly wasting her or her characters' resources to try to find out things about game.
Posted by: Teemu Nurminen | Nov 29, 2003 at 21:02
Ok, that's definitely valid, Raph. However, now we must ask what the net effect of lowering average play time will be. I'd be one to say that the number of subscribers a VW has is related to average play time (time+ => game quality+ => subscribers+). Thus, if the game retains it's quality, it should be tough to lower average play time (though UO show the treadmill to be separate of this). This works against the influx of casual gamers. So, if the goal is more total subscribers, this might not work. If the goal is to open up the genre at the expense of a few subscribers, let's hope the new folks stick with it.
Posted by: Tek | Nov 29, 2003 at 22:39
You asked for additional motivators, and I think Profit could well be an addition; where "Profit" is playing exclusively to gain RL goods/influence.
I think there's a distinct motivational separation in desiring in-game "Achievement," or acquiring things to boost status/power, and the motivation to acquire real-life goods.
It's a very different attitude, I've found, between people who take joy in acquiring in-game items and levels, and those who treat it as a day job, almost as if the game is a chore, a means to an end, rather than an end in and of itself.
-Rip
Posted by: Riprend | Nov 29, 2003 at 23:41
I'm not in favor of throwing out Bartle's Player types in favor of a motivational framework. The typing seems to be a good way of discussing player make up in a more exclusive, braod strokes way (this is not to say that the motivational model is not elegant and suberbly useful). It should be noted that what is being tested here, correct me if I'm mistaken, is motivations and make up of current players, more so even than that, it is a test administered to players dedicated enough that they visit forums and participate in these tests willingly. It must be speculated that the responses given to these surveys are more applicable to core players than casual players (although I wonder if we'll ever know how much). Due to this doesn't it seem to be an slightly improper base for speculation on the best ways to grow MMO population? Especially so if real MMO growth is expected to come from an influx of casual players due to a break in the social stigma of the games?
The survey places an average of what? 24-25 hours per week playing the game?!? Since we assume that the goal of the industry to reduce, actually or illusory, the time in-game (or at least it will be until somebody decides to simply tax player transactions rather than monthly fees) inorder to attract new players and reduce server load there-by maximizing profits this would seem to be an improper basis for detemining what aspects of the MMO should be tweaked inorder to appeal to the greater market since these surveys study primarily 'core gamers who are already snagged hook, line and sinker.
Posted by: Squirrel | Nov 30, 2003 at 02:28
Squirrel - Bartle's "player types" and the motivational framework both try to answer the same thing - "Why do people play?". The only difference is that one is based on analysis of quantitative data and the other is based on speculation.
More importantly, the data analysis shows how Bartle's types are misconstrued and cluster elements that should not be clustered together. See: http://www.nickyee.com/facets/revisiting.html
Although I agree with you that the key to bringing in new players probably doesn't lie in understanding current players because current MMORPGs have become far too unwieldy to casual gamers. But then again, that wasn't the goal of the study. I leave that question to marketers :)
Posted by: Nick Yee | Nov 30, 2003 at 04:12
Nick Yee>The only difference is that one is based on analysis of quantitative data and the other is based on speculation.
My player types model came from a set of observational data, which although not as scientific as your quantitative data is no worse than what a lot of psychologists and anthropologists seem to use. I concede that your results are based on superior evidence, however.
Your results, as you duly note but many of the people who read your conclusions seem to fail to realise, are only as good as the questions you ask and the answers these questions provide. If, for example, you hadn't asked a question about (what you call) immersion, you wouldn't have been able to identify it as a factor. It may be that there are other factors that you haven't yet unearthed because you simply haven't asked about them yet. The factors also come with no labels, so the ones you give to them have to be very carefully chosen; I personally wouldn't call the "immersion" one by that name, for example. It's a cluster for sure, but what does it MEAN? That's the labelling problem.
Of the 6 components you have at the moment, 4 seem to fit my original player types and 2 are what might be termed "meta types". It seems to me that these ("immersion" and "escapism") are steps on the way to why people play, but they don't really seem to be orthogonal to the others. You could, for example, play for escapist reasons and spend your time adventuring or socialising or whatever; role-playing can be used to pretend to be anything - it could be argued that EVERY player is role-playing, it's just that some are more conscious of the fact.
If you were to ask people what they liked to eat the most, you may get some answers of "apples" and some of "oranges" and some of "fruit". A cluster analysis would pick up all three. The fact that (colloquially if not biologically) apples and oranges are both "fruit" would mean that you'd see a cluster where you really ought to be asking a deeper question ("which fruit do you like most?"). It could be that this is what's happening here for all we know: people who answer "escapism" or "role-playing" might when pressed give answers along the lines of the other 4 types. Then again, they might not; indeed, it may be that if you pushed the other 4 people hard enough they could be persuaded to focus on either "escapism" or "role-playing" to the detriment of the other types. It really depends on what theory your cluster analysis is working to.
What we have here is a set of highly useful data that can be interpreted in a number of ways. I'm pleased that my player types paper has led to this degree of statistical analysis being made available, free, to researchers of virtual worlds. If it leads to a superior model of why players play, so much the better. I think this new 6-cluster system is an improvement on we had before, but I also think it has much further to go yet.
Ultimately, the limitations of this approach are that they depend on the questions asked. They can point to flaws in a theory ("where are the explorers?") but they can only propose new theory if you ask the right questions, which means you need the theory in the first place in order to ask the questions (which is why you asked questions about "immersion" and "escapism", right?). What would be really useful would be if you could ask open-ended questions, but then you have a categorisation issue; ultimately, you'd up doing pretty much the same as I did when I originally formulated the player types - looking for themes, then asking follow-up questions to try to nail down what the differences are (if any) between ones that seem related.
I should point out that my model now has 8 player types, rather than 4. I think some of the problems you had with clusters around the original 4 may be more easily resolved with the new ones (particularly the different types of griefer you identified) but others will be harder (2 types of explorer when the data struggles to support even 1). It'd be interesting to see how these pan out in a survey, but you'd have to ask new questions to do that.
Richard
Posted by: Richard Bartle | Nov 30, 2003 at 11:56
Nick: It is evident that the goal of the study is to further the understanding of player motivations (and for Rich to understand the breakdown of player archtypes based on motivations). But to what end is this data to be used? I suppose that primarily it is to further enable us to discuss the makeup of MMO players and their desires. If so to what end is that discussion aimed? I would further suppose that the primary functional output of such a discussion is to develop accurate action-plans for adjusting/creating MMO structures in such a way as to either address deficiences in said strucutres (such as making the gaming orsocial aspects more robust) or to add/improve upon them in such a way as to break the market out of 'core gamers. This is not to say that such a goal is solely a concern for the marketers and the advancement of profit. I would argue that the end game of this discussion is to open up MMO to a larger market (or at least a selective market that is more proportionally representative of the total society) in order to enable MMO to serve as more than just entertainment. Rather, I would guess, that it is in the best interest of all parties to develop MMO so that it can exist as both a profitable and rewarding entertainment/media and as a robust virtual sandbox for the further exploration of socio-political, economic, legal, and philosophical experimentation and excersize.
To wit shouldn't any study of MMO player archtypes and motivations be more closely tied to "Real World" studies (noting, of course, the comparison to the Big 5 included in the motivational study)? I feel as if this avenue is putting the cart before the horse in a sense. That we're studying a nascent and woefully incomplete system with notice paid to "Real World" corellations.
Just a peeve. Perhaps these models should be the beggining of a greater discussion rather than the discussion in and of them selves.
Posted by: Squirrel | Nov 30, 2003 at 13:08
Squirrel: Simply put, I would ask this --
"Why did Newton develop his three laws of motion? Why did Franklin fly a kite in a lightning storm?"
I highly doubt they were thinking about fundamental physics or generating power, but these discoveries were later taken and built upon to become "Real World" applicable.
===
Has anyone tried to reuse Bartle's original method of asking players (or was it wizards?) "What do people want out of a MUD?", after some 13 years? I think it's still safe to say that, among players, a sufficiently significant portion don't know about Bartle's original types. They might turn up something new, unlooked-for, and useful.
Posted by: Michael Chui | Nov 30, 2003 at 14:51
Richard - I actually don't even assume orthogonality in the model to begin with because psychometrics are never orthogonal. Whereas the "player types" seemed to be an orthogonal classification system, the "motivation framework" tries to assess players on all 6 motivations and come up with a holistic way of describing players. But the thing is - without doing any statistical analysis, how do you know the "player types" are or are not orthogonal to begin with?
Richard Bartle >> It could be that this is what's happening here for all we know: people who answer "escapism" or "role-playing" might when pressed give answers along the lines of the other 4 types.
But if that were the case, then "immersion" and "escapism" would correlate highly enough with one of the 4 other compoenents and the factor analysis would have grouped them together. That didn't happen, suggesting that these indeed should be thought of as separate constructs.
Do you have a link to the 8-types model?
Posted by: Nick Yee | Nov 30, 2003 at 16:41
A second thought, while I was reviewing your original analysis. Leadership isn't really a motivation. By the statements listed in that factor, you are told that (1) some people are effective group leaders [with no indication as to whether or not they want to be the group leader], and (2) a lot of people like to be alone.
Also, with your new set of motivations, I, like Richard, noted the similarities between Bartle's Types and your Motivations. Expected. But perhaps there is this to consider: There are reasons people START playing, a different set of reasons people KEEP playing, and a third set of reasons people STOP playing. Obviously, the third set is the hardest to gather, but most designers believe that failure to cater to the second set is the dominant motivation to stop playing.
So you could say people START playing to Escape, but they STAY because they like to Achieve and so on.
Posted by: Michael Chui | Nov 30, 2003 at 17:15
From the very start of my MMORPG days I have found that my and my friend's most common reason for STARTING to play a game is a new innovation, whether it be EQ (which everything was a first in) or AO (where mission terminals and the ability to solo effectively, etc.) or DAOC (realm battles, good newbie crafting system, etc) and on and on. I'm one of those people who LOVE creating new characters, and being able to alter many different parts of my character's abilities greatly enhances MY play experience.
My continuing reason for playing is usually character improvement, which would probably place me as an achiever (I bartle test occasionaly but always get a 80+% Achiever rate). When I'm starting to go downhill is when my progression starts to slow down usually combined with other things.
The reason's I have quit have usually been that I had worn out the new content or that the content didn't offer anything for me to expand upon, or I because tired of attempting to push my way up to a higher tier of the system. OR in some cases where the game kept my interest for a while, the asstacular customer service that seems prevalent in nearly every MMORPG to date (especially SONY/Verant) puts me off with their "no comment" or "it's fixed because we say so" policies.
To hold onto your players, I'd say you're going to have to come up with a way to let characters progress at the rate they did when they were newbies, giving them new abilities or skills all the while and new content in which to efftively pit their newfound skills against. Which would take a hella lot work and lots of code if you ask me :) AO came relatively close I'd say but missions got too repetitive. EVE does the best job to date with "advance like a n00bie" on your 3rd year playing that I've seen yet (and I've seen all your games :D).
It's like a bigjigsaw puzzle but it seems like only parts of the puzzle are put together and the game companies still can't get all the pieces to fit together into the big picture.
Posted by: Lee Delarm | Nov 30, 2003 at 20:49
Nick Yee>the "motivation framework" tries to assess players on all 6 motivations
Well, all 6 that you know about so far.
>how do you know the "player types" are or are not orthogonal to begin with?
Because I have them situated in terms of an orthogonal two-axis model. This doesn't mean that there can't be other axes, but it does mean that the entities I describe in terms of that model are disjoint.
>But if that were the case, then "immersion" and "escapism" would correlate highly enough with one of the 4 other compoenents
Not necessarily: they might correlate with all of them equally well, which wouldn't show up as anything special.
>Do you have a link to the 8-types model?
Yes, but unfortunately it's http://www.amazon.com/exec/obidos/tg/detail/-/0131018167/ .
Richard
Posted by: Richard Bartle | Dec 01, 2003 at 03:09
My problem with this is that the distinctions are not obvious to me. Well, my real problem is that I an not really very keen on this kind of study of human behaviour, so you can of course dismiss my reply on that background, but if it has to be done, at least let's see what is problematic with the typologies:
1) Relationship: Make Friends / Offer and Give Support
Here I miss the development of status, of a personal identity, as well as group dynamics. They are all part of the relationship building that happens in games. I also find that a lot of the aspects under other cathegories belong in relationship.
2) Grief: Scam / Taunt / Annoy / Dominate
All of these are parts of the game not as grief, but as pleasure as well. Grief play is play that drives this part of the playing to a point where it disrupts the game itself, but a good scam, a well-frased taunt, some real annoyance and dominance to fight - all of these are parts of a good gaming experience, as they create tension. I would also put dominate under both relationships and immersion, scam would definitely belong under immersion.
3) Immersion: Story-Telling / Role-Play
This one lacks a lot of playing strategies. First, we can put all of escapism here. A good scam also needs a player who immerses himself in the story of the scam, to make it work. Relationships could also be moved down under this category, as a role-player will have ic and ooc relationships, and the emotional investment in them will be not different in meaning and value, but different in content.
4) Escapism: Escape / Vent / Forget RL Problems
See immersion. But if it is to be treated as its own there should be fascination, intellectual challenge, problem-solving and puzzles.
5) Achievement: Achieve / Accumulate Power
And the challenge, problem-solving and puzzles could be here too. And relationships connect to this, through status: there are different ways of achieving, through points and through social status. You can achieve IC or OOC, or technically.
6) Analyze: Rules / Mechanics / Mapping
Also remember that a player will analyze social relationships, the fictional frame, the history, game politics, language and style.
These are just a few of the important motivations in a multiplayer game. The players also have political, religious and intellectual strategies, spiritual or self-serving both. Motivations for playing are, as far as my research tells me, highly personal and often contradictory.
And an other thing. A lot of people just think playing is fun. Where is the fun part, or is your real project to define "fun"?
Posted by: torill | Dec 01, 2003 at 04:59
Slyfiend said "these days, "roleplay" is generally assumed to mean fabricating a private story with your friends, and playing it out, within the game's mechanics but independant of the game world."
This is so sad, especially because it's so true. It seems to be a direct descendant of the MUD/MUSH style of playing make-believe in a shared imaginary context where the rules and boundaries are defined and enforced by a social consensus (or perhaps a tyranny of the majority) rather than by the game world itself, with occasional assistance from the 'gods'. Done well, it's truly impressive and a lot of fun - like a skilled actor (or troupe) who can stand on an empty stage and pull a whole audience into an alternate reality fleshed out soley through words and body motions. Unfortunately this degree of skill is exceptionally rare and we are far more likely to be treated to the equivalent of the drunken schmuck who tries to hog the stage at the local comedy club on open mike night.
It takes far less skill (and is thus accessible to far more people) to role-play in a setting with props and scenery and atmosphere that support the alternate reality that is being imagined. Proper stage management can provide enormous support with scenery and props that change appropriate to the progress of the story. Also critical is the need for all of the actors present to cooperate, if not in telling the same story, at the very least in the portrayal of the same alternate reality.
Unfortunately, while the current crop of graphical MMOG's certainly provide unparallelled visual stages and props for imaging some of these shared alternate realities that were previously only imagined, that's about all they are good for in terms of roleplay. Well, that and providing the shared stage of course. :) But there are already plenty of shared stages around, including chat rooms, forums, MUDs/MUSHes, virtual worlds, etc., so there's no real advance in that area.
This improved imaging capability has come at quite a steep cost for the role-players though. Forget about incompatible 'visions' of what things should look like - that's no harder to overcome than having an unpainted miniature in the middle of a tabletop RPG battle. Nevermind, even, the restrictions imposed by the rigid game rules and mechanics. While an avid MUSHer may gnash his teeth in frustration that his dark elf can't be a Paladin, or his half-giant warrior can't smash open a simple locked door, at least those rules and mechanics provide a common framework for establishing just what is and isn't possible in that shared alternate reality.
No, the real problems for role-players in current graphical MMOGs come primarly from three sources:
1) The static, pre-scripted, unresponsive nature of the worlds themselves. The stage props and scenery just don't change appropriately with the story. This is not just the inability to alter the physical environment, but the social one is even more critical - the Orc Overlord just won't stay dead, and while you can rescue the Elf Princess, she re-spawns in her dungeon cell before you've even had a chance to get back to the Elf King for your reward! And its even worse when the developers pop in once a year to spend an hour acting out some prescripted story that has absolutely nothing to do with the lives that the players have lived for hundreds of hours up to (and after) that point.
2) The stupidity and obvious mindless robotness of all the NPC's in the gameworld. They're there, it's hard to ignore them, but they just don't seem to be able to roleplay even their extremely limited vending machine or cannon-fodder roles very well, and every time they fail to do so they disrupt the continuity of the shared alternate reality. Not only does this shopkeeper's lines never change, but he doesn't even recognize me and I've been selling him stuff every day for a year now!
3) The unavoidable presence of other players with interfering or incompatible visions of the alternate reality. Not even mentioning those sociopaths who have no vision of an alternate reality at all but are just there to grief others. Mostly this one seems to be a matter of scale - what was easily avoided/policed in a small group of friends, and tolerable or manageable in a moderately sized MUD, has become completely intolerable and unmanageable when scaled up to Massive MOGs.
For these three reasons roleplay in the form of in-character reactions to the world and its inhabitants is a vanishing playstyle. And not only that, but even the 'hardcore' roleplayers capable of sustaining their shared illusion independent of the world and its inhabitants seem to be abandoning either that style of play or the graphical MMOGs that are so inimical to it.
These issues, and the exhorbitant costs of designing, policing, and running games that might attempt to overcome them for hundreds of thousands of players at a time, are why the MMORPG is a thing of the past, if it ever even existed other than in the hopes and dreams of naieve players and developers. What we've got now, and what we are stuck with for the forseeable future are lots of MMOG's, with role-play relegated to single player games or the occasional small-scale player-hosted (and policed!) multiplayer game. Unless and until someone figures out an affordable way to actively support RP and Massive in the same Game.
Posted by: Bryan Allman | Dec 02, 2003 at 11:08
Interestingly enough, a group of Meridian 59 players on a fansite came across Nick's test online and took it.
http://www.gilcon.net/m59forum/topic.asp?TOPIC_ID=2605
Includes my results there, too.
Thought it might interest someone. :)
Posted by: Brian 'Psychochild' Green | Dec 05, 2003 at 05:43