Interview with Dr Richard Bartle: The Bartle Types, Games Studies, and Virtual Worlds as Identity Laboratories

Dr Richard A. Bartle is Honorary Professor of Computer Game Design at the University of Essex, UK.

He is best known for having co-written the first virtual world MUD (“Multi-User  Dungeon”) in 1978, and for his many contributions to all aspects of virtual world design and development. His 1996 paper “Players Who Suit MUDs” established the first model of player archetypes in virtual worlds and has seen widespread adoption within the video games industry. His 2003 book “Designing Virtual Worlds” is the standard text on designing virtual worlds.

In 2010, Dr Bartle was the first recipient of the prestigious GDCOnline Game Legend award.

His most recent book, “How to be a God” (2023), discusses what the design of virtual worlds tells us about our own reality, answering philosophical and theological questions, but raising yet more.

In our interview, we talked about how working on MUD piqued his interest to look deeper into the motivations of players, and how that eventually led to one of the foundational papers in the field of virtual worlds.

The Bartle Types, Games Studies, and Virtual Worlds as Identity Laboratories

Dirk Songuer: Richard, thank you so much for taking the time, I really appreciate it.

Richard Bartle: Hello! So, what is it you want to know?

Dirk Songuer: You created MUD together with Roy Trubshaw in the late 1970s. I am interested in what happened next, specifically when you realized that this experiment-turned-game was also interesting from an academic point of view.

Richard Bartle: When we started with MUD, it wasn’t done as a proof of concept. It was done because we wanted to create a world. Or rather, there was a proof of concept that Roy built the very first day just to test the technology. After that, he immediately began working on the first full working version that would be recognizable today as a MUD – that was in assembly language – and then there was a third version, which we wrote in BCPL. That version became known as MUD1.

Now, looking at it academically, back then games weren’t academically respectable. So, you couldn’t look at MUDs academically because you would never get published, you wouldn’t get funding, you wouldn’t get anything.

I stayed on to do my PhD in Artificial Intelligence because AI was interesting. Essex University was one of only three in the UK that had kept its AI group once funding for academic AI research was withdrawn. The reason I chose it, as opposed to any of the other areas I could have gone into, was that I wanted to have the non-player characters (NPCs) in MUD1 act clever. I wanted them to be able to think.

I couldn’t use MUD1 in the PhD, because too few people knew what it was. Instead, I did the PhD thesis with a playing card game, a patience variant called Scorpion. There are two ways you can approach it. One of them is asking: “What do I want to do, and how do I figure out how to do it?”, and another one is: “What can I do, and how what the consequences of doing it?” One is backward-chaining; one is forward-chaining.

Most of this was a part of AI call “planning”, which is now “good, old-fashioned AI”, because it’s all symbolic rather than statistical.

My PhD allowed me to switch between backward- and forward-chaining by making planning actions first-class objects, which was not something that planners could do at the time. I called that “cross-level planning” and later implemented it in MUD2. I gave it to a few of the non-player characters and it worked well. The trouble was that the system was too smart. I had to dumb it down a bit so that the players weren’t taken down by relatively simple monsters.

Anyway, that’s why I did AI and I’ve been interested in it since then.

Dirk Songuer: Throughout all of this you were still running MUD1 and working on MUD2?

Richard Bartle: That’s why Roy and I set up MUSE Limited (Multi-User Entertainment Ltd.). Roy worked on the on the account system and the underlying physics, while I worked on the gameplay.

Dirk Songuer: When did you get interested in player motivations?

Richard Bartle: Well, game designers are interested in many things. I was creating board games and if I wanted to make a game about pirates in the Caribbean, I’d read a book on pirates in the Caribbean.

It was through running MUD1 that I noticed things about the players. We had this discussion amongst the “wizards” and “witches” – the senior players who had won the game and had been elevated to have administrative privileges.

Some of them liked being visible to everybody. Like a performance, if you like. “I am the God of this world, and I can do all these things!” Others liked to stay in the background, invisible, so that nobody knew.

There might have been something like six wizards and witches in the game at a given time. One of them goes visible and everyone wants to say “Hi!”, thinking that there’s only one there. Suddenly, all the little nuanced changes they were making were gone, because this one wizard or witch became visible and started attracting all the attention.

This led to an argument between the wizards and witches: “What is it that we’re meant to do? How are we supposed to make things fun for the players?

Dirk Songuer: To clarify: When a player reached the maximum level, they automatically became a wizard or witch. And as such, they were expected to create or facilitate experiences for the players and establish a governance structure, correct?

Richard Bartle: Generally, governance went quite well because the people knew that we were watching them. “Who watches the watchers?” Well, we did, the arch-wizards.

We didn’t have to do much, but if we hadn’t been there, we would have had to do a lot more. In essence we were needed in order to let the people know that there were greater powers than them.

Honestly, though, most of the people that made it to the end game just sat around, saw the powers they had, said hello, and then then left quite quickly. In a way, they had finished their journey. Others wanted to hang around to stay in touch with their friends.

Dirk Songuer: Their argument was caused by having different ideas what it meant to be a wizard or witch. Given that they wielded God-like powers in the game, didn’t that also lead to problems?

Richard Bartle: Well, such a system doesn’t scale up, let’s put it that way.

We could operate like that because we had a small number of players, only a couple of hundred players, maybe 300. Today’s virtual worlds have many, many more players than that. Maybe you could try such a governance structure on a server-by-server basis. If you do that, you end up with what we already have in player-organized structures: Guilds.

If you have a player guild with 200 members, then the hierarchy of the people in charge of the guild are doing what the wizards and witches used to do in the past – with some powers over other players. The powers of guild leaders aren’t quite as strong, but they still have the ability to kick you out of the guild. We could create a service around this structure, where everybody on the server is automatically a member of a guild, and the guild management sets the rules. If you don’t like it, or they don’t like you, then you can quit and go to another one.

That means that instead of having 10 servers with 10,000 people each, you have 1,000 servers with 100 people playing on each one. You’d have more servers with fewer people on them, and that’s the old MUD model.

Dirk Songuer: Besides governance, what was the conclusion to the argument? Did they agree on how to make things fun for players?

The question was raised: “What do players find fun?”. In the discussion that followed, people brought up a lot of things that they found fun, or what they used to find fun but not anymore, and so on. After a while, the conversation petered out. Towards the end I thought: “Well I’ll better summarize this.

I was categorizing what people said and on the surface it was more or less in tune with what I’d previously thought. Analyzing their answers further, there seemed to be four archetypes – but why four? So that’s when I got to the graph with passive and active, players and world.

In the earliest model of player types, Dr Richard Bartle identified two main sources of fun for players: The game itself, including the narrative, mechanics, user interface, and surrounding elements. The other source being players within the game, the direct and indirect interactions and relationships they have with each other.

He also observed that players took different approaches to experience fun. Some players preferred an active playing style, using the systems and mechanics provided by the game to do things. Other players preferred a passive playing style, wanting to be entertained by the game play, narrative, and systems.

The inter-relationship of these two dimensions can be represented as a graph: People being active within the game or when interacting with other players, and people passively experiencing the game or in their interaction with other players:

  • Active + game-oriented = “People acting within the game context to achieve game-related goals,” or “Achievers”
  • Active + people-oriented = “People acting upon others to manipulate or cause distress,” or “Killers”
  • Passive + world-oriented = “People experiencing the game world as much as they can,” or “Explorers”
  • Passive + people-oriented = “People having experiences with others, utilizing the game’s communicative facilities,” or “Socializers”

Later, Bartle conceded that the “active / passive” dimension was misleading, as exploration and socializing was clearly an activity. Based on input by Alan Schwartz, the founding editor of the Journal of MUD Research, he changed the axis to “action” and “interaction”.

Dirk Songuer: These notes directly led to the article in Comms! Plus magazine?

Richard Bartle: Even before the Comms Plus! article (“Who plays MUAs?”) I noticed things about the players that later became player types. There’s an earlier article in the Member’s Dossier of the “Adventurers Club Limited” in which I talked about player archetypes. They didn’t have all the same names as they do now, but I already had the idea in my head just from watching what people were doing.

Eventually, I sent the article to Comms Plus! because that was the only magazine that was really devoted to virtual worlds at the time. The editor wanted more people to buy or subscribe to the magazine and he would say or allow controversial articles to try and get conversations going. Maybe he was a bit of a troll in that respect.

Anyway, the article didn’t go down too well so I thought “Let’s leave it for a bit.”, but I kept using the model for MUD2. I followed what the players were doing and how they fitted into the model; Then, I designed content for particular player types – although as it happened, I was already doing it by instinct. I was putting things in the right place for the players to find when they needed them.

Dirk Songuer: And later you turned your observations and earlier article into a scientific paper for the Journal of MUD Research.

Richard Bartle: Well, I wasn’t writing it to further my academic career. I saw there was a new academic journal coming out and I knew how to write academic papers; I thought “This needs help” and I wrote it all up for that.

Dirk Songuer: Although it is based on the article for Comms! Plus, you made significant changes for the scientific paper.

Richard Bartle: Yes, I was trying to be clear. I was speaking to game designers, but I knew that it was a formal journal. I needed to explain the model in a manner that was coherent, not just for designers.

I was hoping that other researchers might pick it up and take it further, do better things with it: then, we’d get better models, and if you get better models then we get better virtual worlds. That’s what I wanted.

Dirk Songuer: A big change was the expansion of the core assumption. In the Comms! Plus article you started from the assumption that “People play MUDs to have fun.” In the paper you expanded this into a categorical statement that “The fundamental purpose of virtual worlds is the celebration of identity.

Richard Bartle: Yes, the individual levels of “fun” are what you need in the moment to progress your sense of who you are.

You’ve got the four player types, but people don’t stay in those, they drift. They move between them in certain, observable ways, and that’s why I eventually added an extra dimension, making it eight types. It enabled me to explain how this progression was working.

Essentially, you start off by finding out where the boundaries are; then, you spend time figuring out what you can do within those boundaries. Then you play within the boundaries to reach the goal that you believe the game is about, and in the end, you hang around with your friends or just explore the world on your own. There’s no drive except your own internal one.

That was the path that people were following, which is really a path to self-identity. Some academics had already noticed that players were entering virtual worlds in order to figure out who they were in that world. I didn’t know until I was doing research for the book “Designing Virtual Worlds”. I did know some of the papers, but Google wasn’t a thing back then, and it was hard to find the relevant research.

The original player types weren’t there to say: “These are all the player types.” I wasn’t expecting that the archetypes were going to be definitive. I just wanted to say “Look, here are some other types, and people are playing for these different kinds of fun.” It was about: “There are more player types than just you.” That was the message that I wanted to get out, but it seems I hit on player types that were actually accurate.

Dirk Songuer: If identity exploration is the fundamental goal of virtual worlds, doesn’t that mean that virtual worlds by definition need to be a little bit uncomfortable? Or rather, on the edge of what you are comfortable with as you explore edge cases of yourself?

Richard Bartle: Well, yes, but they could be comfortable. People can feel uncomfortable in their real life, but as soon as they’re entering a virtual world, they feel comfortable. I’m not saying everybody does. I’m just saying that it’s not always the case that you have to feel uncomfortable if you go into a virtual world.

You might be taking some uncomfortable steps that you wouldn’t do in real life, for example decisions in a game world might be different to decisions in real life. On the other hand, you could go into a virtual world and just avoid all the uncomfortable things, and still find yourself being channelled towards the person you really are.

The thing about today’s virtual worlds is: It’s very difficult to play them and not take too much reality in. If you’re playing in a virtual world, chances are that you’ll be talking to people over Discord or whatever. The fact that you’re speaking, with your real voice, means that you’re still attached to the real world. It’s very difficult for you to be someone you’re not if you are the one still speaking.

The reason you’re speaking with your real voice is because console players are also entering these virtual worlds, and they can’t type. So, they have to use their real voice to communicate. This won’t change until we’ve got proper voice fonts.

I’m speaking to you now in my northern English accent. I might not be able to change the accent, but what if I want to sound deeper and cool? I might want to sound female, or more elvish, or orcish, whatever that means. I could design and tune my voice myself, or maybe buy a voice off the Internet. If I don’t sound like myself, I can wear a different identity – and that allows me to explore in the virtual world who I am through play.

Dirk Songuer: Now we have virtual worlds on consoles, computers, and even mobile phones. And it’s not just the platforms people use to enter the worlds. While MUD was the first virtual world, it spawned an entire ecosystem practically overnight. Suddenly there were many types of virtual worlds, with different themes, some not being games at all. In your 2003 book “Designing Virtual Worlds” you organized this into neat periods and generations. But really, this was all happening all at once.

Richard Bartle: There were about six initial games that people wrote from scratch, including MUD1, then there were the ones that people wrote based on those games, and then there was a big split between all the social worlds, like MUCKs, MUSHes, MOOs, TinyMUDs, and the game worlds like DikuMUDs and LPMuds.

There were at least a couple of thousand DikuMUDs at one point. We also had graphical worlds at the time. It was graphics in text, but still graphics. There was Kesmai using ASCII characters. There was Avatar on the PLATO system, which used tiny little line drawings.

Just before the Internet went commercial maybe 11% of the entire Internet traffic belonged to MUDs.

Dirk Songuer: Didn’t that popularity help to make virtual worlds academically-respectable?

Richard Bartle: They still aren’t academically-respectable – not as much as they could be, not to the extent that, say, film is.

Most of the academic research was conducted in social worlds because, again, games weren’t academically-respectable. As a result, the people who wanted to study virtual worlds didn’t use game worlds. They went and studied virtual worlds that eschewed game mechanics. LambdaMOO was the poster child, but it wasn’t the only one.

Dirk Songuer: But there was academic research happening?

Richard Bartle: As soon as there’s a new area to research, people will descend upon it. In the early days, many players were going to be academics anyway. The big-enough timesharing machines that you needed to run virtual worlds were in academic environments. A lot of these machines were at universities and students would play games on them. Then others would come and look at what people were doing in these games and suddenly you’ve got a research topic.

In the real world, the commercial world, marketers were interested as well. They wanted to know who’s playing, why they’re playing, and how to get them to play more.

We had psychologists looking at virtual worlds; theatre and literature studies; anthropologists had a look and wrote some books, then the lawyers, who stuck around for a lot longer. We had feminists come along to decide whether to be in favor of it or not.

There had been some people doing term papers, basically as a way to play games instead of doing actual academic work. In the end they would send out some questionnaires and they would write it up: This percentage male, this percentage female, and they play for this long, that sort of thing.

We had geographers and architects look at virtual worlds at some point, but they mainly looked at graphical worlds like Second Life and were largely unimpressed.

Some of this research was interesting, but most of it was very superficial. Multiple disciplines came in and attempted to fit virtual worlds into their particular world view. Once they’d done it, they left.

Then there were people like the “Think of the Children!” brigade. It doesn’t matter what the game is, they argue it’s really about you wanting to cause pain to others. They still don’t get virtual worlds, they really don’t. When you look at the level of research that they do, it’s staggeringly bad and easy to refute.

So not all academic work is good. Nevertheless, there are some decent researchers and journalists out there. Of course, there are Julian Dibbell’s articles, and they’re still relevant – in fact, I just discussed them with my students last week.

Nick Yee’s work about motivation in virtual worlds is really interesting, and now he’s expanding into motivation in everything games, even board games. I have a lot of respect for what Nick is doing.

Ultimately, back then there wasn’t a lot of interest. Virtual worlds hadn’t pricked the real-world consciousness enough.

Dirk Songuer: Why do you think that was?

Richard Bartle: The thing is, people have their own focuses. Think about the amount of verbiage written about designing games for female players. Don’t get me wrong, that’s great because we need more women in games, but Nick Yee’s research for example showed that age is a much bigger factor than gender when it comes to play – but nobody is studying age in virtual worlds because no-one is interested in how older people play games. Instead, they’re interested in how people just like them play games. Also, gender is a relatively easy division, whereas age? Where do we even start?

So we get some things that are over-analyzed, and some things that aren’t analyzed at all when they should be.

The problem is that academia is divided into silos. Each individual field only looks into their own thing, but the issues raised in virtual worlds are cross-disciplinary. Academics will say that cross-disciplinary research is good, and indeed it is good, but there’s not always funding for it – and certainly not when it involves games. I’ve done cross-disciplinary work, but it was framed as “natural language processing,” not as “text-based virtual worlds”.

When does something not-real count as being real?” “What can you do in a virtual world that you can’t do in real life?” “How much protection does a virtual world need to give you?”What precautions must you take if you want to create an experience like this?” All these things need examining, and people do come along and raise these questions, but they never give the answers.

Dirk Songuer: These are almost questions about the nature of virtuality, not just about MUDs.

Richard Bartle: There are people who think: “It’s just about text-based MUDs and now we’ve got graphics. That must be completely different!” No!

The same people who think that would themselves be swept aside by the next iteration of technology. “Oh, that’s just 2D graphics, we’ve got 3D now! We got VR now! It’s the Metaverse now! None of that applies!” In reality, you’ve just made minor changes in the interface. The worlds themselves are the same.

How you structure these virtual worlds, and how people interact with them, that is very interface dependent. You would put different things in the world depending on the interface – but behind the scenes, they’re all the same. It’s as if text-based MUDs are silent movies, and graphical MUDs are movies with sound, and then VR is color movies, but overall, most things still apply. Most of the old techniques do still work and are still the same.

Dirk Songuer: That said, there are big differences in the way you experience the virtual world through the interface, and how they achieve a feeling of presence. In a way, text-based virtual worlds use the mind and imagination as immersive technology, supported by the computer, but not rendered by it.

Richard Bartle: That’s right, the virtual world is maintained in your head. Like the real world, everything you see goes into your head and makes you create a cognitive representation of it. If it’s a textual world then that representation relies on your imagination. The graphical world will give you a lot more visual information.

The downside is that you have less opportunity to interpret the world yourself. With a textual world you can interpret it differently, but then the same world can create different images for different people.

Another thing about text is that the input and the output are within the same modality. In a graphical world, what I’m typing is one modality and what I’m seeing is another modality. I have to switch between words and pictures. That means you have to keep switching, even if you do it on a subconscious level. With text it’s all the same.

The problem with text-based worlds is that you must be able to read and write. It requires imagination, and not everybody has an imagination strong enough – or they do, but they don’t exercise it in that fashion. That’s a downside, particularly in an industry that’s trying to lower the barrier of entry for newcomers.

On the other hand, the industry also wants to raise the bar by increasing the fidelity of the graphics. Back in the 1990s, people were already talking about realistic graphics: “Gosh, if the graphics are this good now, imagine what the games going to be like!” Well, they were pretty bad, because they’d spent all the money on the graphics. If you train people that “cutting edge graphics” is what defines games then people will be disappointed if they see text.

Dirk Songuer: And even though a new modality might bring benefits, it’s easy to forget what other benefits the replaced modality had. For example, you can have sensory experiences like smell, taste, and touch through the primary interface of text without breaking immersion. This is not necessarily true for graphical interfaces – yet.

Richard Bartle: I make all my third-year students play MUD2. Since text has fallen so out of fashion, many haven’t come across it before. Amazing things happen when they start to play:

“What do I do here?”

Well, what do you want do?

“I want to get the book off the bookcase!”

Type it in: Get book from bookcase.”

“Oh? Oh, wow!”

By contrast, games today have very limited options. In MUD2 they could do it all because possible actions don’t have to be represented by some combination of buttons on a controller, or a button they can click on. Text has a much wider scope for functionality. Also, you can create nuanced differences between things. For example, the difference between kicking things, hitting things, and biting things, rather than just an attack button.

Dirk Songuer: If virtual worlds today are more limited in terms of verbs and nuances, and the focus has shifted towards immersion and graphical fidelity, maybe the research focus also shifted towards these topics?

Richard Bartle: A lot of virtual world research is old and talks about text-based MUDs, but it still raises fundamental questions: “What’s real, what’s virtual? What happens when they interact? What overlaps?

We used to have Usenet groups and mailing lists where most of the designers and developers of virtual worlds would converse. A lot of the issues were thrashed out there. By the time the graphical worlds came out, we had a blog called Terra Nova.

Every once in a while, tech people come along and assume that just because their technology is new, none of what was true in the past continues to apply. Maybe they don’t look at past things because then they would have to admit that they were ignorant, and that they’re not at the forefront. I’m not saying they’re cynical about this. It’s just that this is how it worked out. Most people coming into the area of virtual worlds simply aren’t aware of the already-existing work. I keep saying: “Look! The answers to these are on these platforms! Have you looked at this? Here it is!

Nowadays of course, no one is going to look at Terra Nova, even though it’s a gold mine for this sort of material. They’ll argue that: “Oh well, the Metaverse is different! This is VR, 3D now! Let’s look at all this from a different perspective! People are now sensorily-immersed in the virtual world!

First of all, that doesn’t mean anything. I’m sensorially immersed in the real world, but that doesn’t mean it’s necessarily better than anything virtual. Otherwise, I wouldn’t play game worlds. They try to ignore all the learnings of the past and so repeat the old mistakes all over again. We could save a lot of very rich people a lot of their money if they just asked about the past.

Dirk Songuer: When it comes to your work around fun in virtual worlds, what should they ask? What do you think is the most misunderstood aspect of your work?

Richard Bartle: The most misunderstood part of my work comes from people that look at player types, but don’t read the paper. People often think that player types apply to games in general. The paper was meant for virtual world developers, but some developers of ordinary games also latched onto it. The model might apply to games in general, but I make no claims that it does, because I can’t explain why it would. So, when people try player types in games and they don’t work, then I get the blame – but I never said they would work for everything. If you’re using them for some kind of mobile phone game, well, no, they won’t work. Player types have been wrongly applied to a lot of things. Gamification used it, but they never asked if it makes sense.

Another thing is that people decided that the “Killer” archetype are people who like Player-versus-Player (PvP) mechanics. No, you misunderstood that completely. “Killers” might like PvP mechanics – if they can gank someone, grief them, kill them when they weren’t expecting it, or kill others that don’t like to participate in PvP, kill people that are begging them to stop. That’s where they got the name from, but they’re not just people who like to play Call of Duty together.

Sometimes people build virtual worlds around the concept of player types and say: “Oh, we extended the player-types model!” What they mean is: “We broke the model because we didn’t understand it.

Dirk Songuer: The original article in Comms! Plus can out in in October 1990, over 30 years ago. The paper in the Journal of MUD Research was published in 1996. Looking back, what do you think about them now?

Richard Bartle: There is a lot of clarity in that paper. On the whole, there’s not a lot that I would disagree with today. Obviously, it’s very dated because it’s close to 30 years old by now, but just because it’s dated doesn’t mean that the inherent truths of it don’t still hold up.

When I wrote the paper, I wasn’t thinking that “This is going to start the field of Game Studies”, but it turned out to be one of the foundational papers.

I stayed on as an academic because I was on a conveyor belt, and I wanted to ride it to the end. So, I wound up with a PhD and lectureship, now a professorship, but I wasn’t trying to put virtual worlds into academia. A tree isn’t trying to get its acorns to be eaten by birds. That just happens. You create all this content, which you hope is going to flower, and other people come along and peck at it. Then they take away your ideas and do their own thing with it, and that’s kind of what happened.

Roy and I lit a fire. We developed a virtual world because we wanted it to spread, and the fire did spread. The whole point of virtual worlds is that you can make these things what you want of them.

Don’t just listen to Roy and me. Create what you want. Make your own version, or interpretation, or alternative to reality. This is what virtual worlds were meant to be.

2 responses to “Interview with Dr Richard Bartle: The Bartle Types, Games Studies, and Virtual Worlds as Identity Laboratories”

  1. Richard supervised my MSc thesis at Essex Uni in 1985. He was a great supervisor, friendly and approachable. Although we knew of MUD, we just thought it was a little quirk of his. We didn’t realise it was the first cyber world. We didn’t have enough login time on the CS computer to spend that precious resource playing MUD. We had to use it to complete our assignments. It must have been research students and academics playing MUD.
    I’m now an AI professor in New Zealand, so Richard had some influence.

  2. I pretty much agree with everything that Richard says, except that my primary motivation – beyond enjoying solving the implementation problems – was to create a world where magic worked and cool stuff happened that was only limited by our collective imagination.

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.