This post stems from two sources. On one hand, Daniel Pink’s Drive has brought variations of self-determination theory of motivation into the mainstream. On the other hand, everyone in the gamification scene is building varieties of Bartle player types to explain motivation. Can these two be brought together?
Actually, Andrzej Marczewski has already done something like that with his user types, but in this post I want to dig a bit deeper into the theoretical basis of doing so. We’ll get back to the user types later, but first we need to venture into Nick Yee’s well-known paper, Motivations of Play in MMORPGs: Results from a Factor Analytic Approach.
First, a short detour into intrinsic motivation. In Drive, Daniel Pink wrote of three factors: autonomy, mastery, and purpose. On the other hand, self-determination theory (popularization of which Pink’s book is) posits three factors as well: autonomy, competence, and relatedness. Not an exact match there, as although mastery and competence are pretty much the same, purpose and relatedness are not.
Self-determination theory does mention life goals as well, as one of the factors that produce individual differences, and those match purpose quite nicely. Relatedness is not taken into account in Pink’s book.
Combining these sources results in intrinsic motivation coming from four factors:
Autonomy. People want to have control over their life and their own activities.
Mastery. People want to get better at what they do and control outcomes.
Relatedness. People want to feel connected to other people.
Purpose. People want to be part of something greater than themselves.
Let’s keep the four of these in mind, and venture into Bartle player types and Nick Yee’s research.
Bartle player types
Richard Bartle described four player types based on his observations of early MUDs in the late 1980s. He attempted to answer the question, “What do people want out of a MUD?” The player types he came up with are:
- Achievers: Players who set themselves game-related goals and set out to achieve them.
- Explorers: Players who explore the game world and its limitations.
- Socializers: Players who enjoy interacting with their fellows, for example through role-playing.
- Killers: Players who try to hurt other people in the game.
Bartle player types have become an often used tool in the game development toolbox.
Nick Yee’s motivations of play
Nick Yee conducted an extensive survey of MMORPG players, in which 3200 players answered 39 multiple-choice questions. The results were then analyzed through factor analysis and grouped into subcomponents, which were further analyzed into main components Overall, 10 subcomponents were discovered, which in turn belonged to 3 main components.
The main components were independent of each other, i.e. whether players were motivated by one of the main components had no effect on whether they were motivated by the other main components as well.
The discovered main components were named Achievement, Social, and Immersion, and the subcomponents Advancement, Mechanics, Competition, Socializing, Relationship, Teamwork, Discovery, Role-Play, Customization, and Escapism. How would these components map into the types of motivation proposed by the self-determination theory?
- Advancement – achievement: progress, power, accumulation status. This maps into mastery.
- Advancement – mechanics: numbers, optimization, templating, analysis. This maps into mastery.
- Advancement – competition: challenging others, provocation, domination. This maps into mastery.
- Social – socializing: casual chat, helping others, making friends. This maps into relatedness.
- Social – relationship: personal, self-disclosure, find and give support. This maps into relatedness.
- Social – teamwork: collaboration, groups, group achievements. This maps into relatedness.
- Immersion – discovery: exploration, lore, finding hidden things. This maps into autonomy.
- Immersion – role-playing: storyline, character history, roles, fantasy. This maps into autonomy.
- Immersion – customization: appearances, accessories, style, color schemes. This maps into autonomy.
- Immersion – escapism : relax, escape from RL, avoid RL problems. This maps into autonomy.
OK, I am sure some of you will want to debate the accuracy of these connections, but to me they mostly seem clear-cut. It seems that Yee’s motivational components can be mapped directly into groupings from self-determination theory: autonomy, mastery, and relatedness!
Furthermore, it seems that the immersion component is actually a misnomer! In particular, the discovery subcomponent is not a very good fit for immersion, but it is a good fit for autonomy. Likewise with customization, where it is not necessary to be immersed in the game to enjoy customization when it is understood from an autonomy point of view. Immersion should be relabeled as autonomy. This also means that Bartle’s interpretation of immersion as motivation that is uncovered through play seems incorrect.
Advancement and social are better matches, and replacing them with mastery and relatedness is simply replacing one word with the same meaning with another.
Umm, wait! What happened to purpose? Remember how purpose in the actual self-determination theory (as contrasted to Pink’s popularization) was in the form of life goals, one of the individual differences? If purpose is a general class that varies from person to person, it cannot be uncovered as such through factor analysis. Purpose is behind all 10 subcomponents, but it is impossible for it to show up as a component of its own in this type of study.
What is quite intriguing about this is that if my interpretation of Yee’s data is correct, Yee’s data actually lends further support to the self-determination theory of motivation.
Revisiting Bartle types
Yee pointed out multiple ways in which Bartle’s player types conflicted with the data Yee had gathered. With self-determination theory as our background, we can also see more clearly why these conflicts happened. Here are the four types of conflict Yee described seen through the lens of self-determination theory:
- Socializing and role-playing are not the same unlike in Bartle player types. These are independent motivations, because one is based on relatedness, and the other is based on autonomy.
- Achieving and competing are linked unlike in Bartle player types. This links comes from both being highly influenced by mastery.
- Exploring the world and tinkering with the system are different unlike in Bartle player types. Exploring is part of autonomy, whereas tinkering is part of mastery.
- Immersion components do not exist in Bartle types. I don’t think self-determination theory helps explain this gap. However, I think it points to something far more interesting. The reason customization and escapism are absent from Bartle player types is that these motivations did not exist in his original focus group. The early MUDs were not very good for customization and the player base was very limited and not prone to escapism. This, in turn, implies that the player demographics and motivations can change over time and research can become obsolete. It also implies that it should be possible to supercede Bartle player types already, because motivations not covered by them have been verified.
There are further implications here. Bartle player types do not cleanly map into motivational factors. They are in fact compounds of various motivational factors. This, in turn, means that Bartle player types are useful insofar as they capture major groups of players who share similar groups of motivational factors, but they will also inevitably fail to grasp many nuances, such as the ones outlined by Yee in his four points of critique.
Mapping the players into four types along acting-interacting and players-world axes inevitably leaves out many details. All models that attempt this same four-field method will be incomplete. Nonetheless, this is different from them being useless, just like Newtonian physics is still not useless even though we have more accurate models at our disposal. A more complex model would need to be simple enough to actually use, if it were to replace Bartle player types.
Marczewski’s user types
Andrzej Marczewski has attempted to build a classification of users similar to Bartle player types for gamification purposes. He describes eight user type, four of whom are intrinsically motivated and four extrinsically motivated. Here I will focus on the four intrinsically motivated user types, because they are a good example of the shortcomings when attempting to map Bartle player types into motivational factors one by one.
Marczewski uses a similar four-field, with acting-interacting and users-system axes, very similar to Bartle. His four intrinsically motivated user types are:
- Acting – users – Philantropists: Motivated by purpose, want to give back to others.
- Acting – system – Achievers: Motivated by mastery, want to be the best at things.
- Interacting – system – Free spirits: Motivated by autonomy, want to explore and create personal content.
- Interacting – users – Socializers: Motivated by relatedness, want to interact with others.
Self-determination theory and Yee’s research point out a significant issue in this classification: purpose is not a uniform factor, different people have different goals. In this type of four-field, the philantropists could just as easily be replaced with competitors, who also act on people, but their motivation could just as well be mastery. But that would ruin the one-to-one mapping from these axes to motivational factors! Exactly. It would ruin that mapping, because that mapping does not correspond with reality in the first place.
It is highly attractive to map Bartle player types into motivational factors. Hey, I immediately tried it myself. Unfortunately, it just does not work out, because as Yee’s research shows, Bartle player types are compounds of motivational factors. Acting-interacting and users-system axes do not form distinct motivational factors.
Towards an improved model
It is important to note (as Bartle does in this speech) that Bartle player types, as well as Marczewski’s user types, strive to be actionable tools for a designer. Yee’s components do not. How do you design a game for escapists, anyway?
Some translation from the motivational components to the design arch-types is needed and even desirable. The question is how to build a model that is both usable and as accurate as possible.
Problem is that there are just many axes on which to model the system, for example:
As a little experiment, what if we used Yee’s three main components, which we know to be independent of each other, and thus can appear in any combinations, and created eight player/user types based on low or high values in each? I’ll borrow some terminology from Marczewski for a couple of these that are equivalent to his user types.
- High mastery, high relatedness, high autonomy: Hardcore fan
- High mastery, high relatedness, low autonomy: Competitive team player
- High mastery, low relatedness, low autonomy: Achiever (possibly PvP’er or griefer)
- High mastery, low relatedness, high autonomy: Explorer
- Low mastery, high relatedness, high autonomy: Role-player / Party star
- Low mastery, high relatedness, low autonomy: Socializer
- Low mastery, low relatedness, high autonomy: Free spirit
- Low mastery, low relatedness, low autonomy: Extrinsically motivated! (probably a gold farmer)
By the way, do you notice how this model solves the criticism directed at Bartle by Yee, as it can cope with multiple motivational factors influencing an individual at the same time.
Compared to Marczewski, while this is an 8-type classification, the axis are somewhat different, as Marczewski’s model is based on only one intrinsic motivational factor being the main one and the third axis is extrinsic motivation, which in this model is one subtype, the one where all intrinsic motivational factors are low. As a result, there are three exactly same classifications, while purpose is missing from this model, and combination of intrinsic and extrinsic motivation are not included either. Marczewski’s model, on the other hand, is missing out on combinations of intrinsic motivational factors.
One clear issue is that this model is missing purpose as a motivational factor. Then again, purpose is very difficult to model, as it varies between individuals, and I am not convinced that Marczweski’s model where purpose is relegated to a single altruistic notion, is any better in this regard.
Obviously, this is but a rough sketch, but I think this might be one way to improve the player type theory, if developed further.
Photo: Gamers by RockaSalvatella @ Flickr (CC)