Over the years, I’ve looked into so many different aspects of online gaming that it’s become harder and harder to think of new themes to explore using a survey format. Oftentimes, I’ll think of a great idea and then realize that it was something I already looked into a couple years back. A lot of the areas that I could move into would require a much more significant effort and devotion of time. These include looking more rigorously into changes over time or delving deeper into many of the more qualitative issues. But unfortunately, time is harder and harder to come by now that I’m working full time.

Getting my PhD also meant being more disciplined about what I say and put out, and this had the effect over the years of putting more and more time into each issue of the Daedalus Project. As a personal project, the time it takes to run Daedalus began slowly to directly compete with time I could spend writing an academic paper among other things.

I’m still deeply interested and fascinated by the MMO space, but I think I came to realize that continuing the project would require more time than I could devote to it. It’s for these reasons that I’m putting Daedalus into hibernation mode. Of course, all the data that has accumulated here will remain and comments will be left open, but there will no longer be regular data updates or surveys.

I look back over these 10 years of doing these online surveys and I’m deeply appreciative of every gamer who filled out my surveys (and there are over 70,000 of you who did), every person who sent me notes of support and congratulations, and especially those of you who have been participating and commenting on the project since my undergrad days and have helped spread the word in the gaming community. I really do mean it when I say that I couldn’t have done it without you. I hope none of you think of the Daedalus Project as something I put together alone because it was something that we came together as a community and built over the course of 10 years.

To all of you who have been a part of this project, you have my sincere thanks and gratitude. And to all of you who venture into Azeroth, Norrath, and other lands, may your journeys be safe and your adventures bountiful!

Kool Aid in Goggles

In the spring of 2005, I was taking a cultural studies course that traced the development of personal computing from the Cold War era through the Counterculture era. Virtual reality (VR) intersected with the Counterculture in an interesting way. During the countercultural movement, the fascination with technology that triggered feelings and sensations of global consciousness drove the appeal of strobe lights, Day-Glo paint, LSD, and also virtual reality systems. What was special about VR was that it allowed people to become disembodied in a space that could be distorted at will. As Barlow wrote after his first experience in virtual reality, “Suddenly I don’t have a body anymore … the closest analog to virtual reality in my experience is psychedelic”.

What was interesting was that this historical theme seemed so out of place in the context of contemporary virtual worlds where embodiment was the unquestioned status quo. Whether in Second Life or World of Warcraft, it would probably baffle most gamers to ask them what a virtual world would look like without bodies? Or what such a world would be used for? But in the same way that games like World of Goo and Crayon Physics Deluxe challenge the status quo of gaming genres, it’s important to keep in mind that the virtual worlds we currently have may not encompass what virtual worlds can be.

====

Are People still People when they MOO?

My first encounter with a virtual world was actually LambdaMOO. In my junior year in college, I was in a small seminar with Doug Davis (who later became my thesis adviser) that looked at the intersection of Freud and the Internet. Although strange at first glance, the hyper-textual nature of Freud’s Interpretation of Dreams makes sense once you flip through the book and see the quantity of footnotes and pointers to other parts of the book. In the seminar, after reading Sherry Turkle’s Life on the Screen, we had an assignment to try out a textual MUD/MOO ourselves. And so, one night in the late winter of 2000, I logged onto LambdaMOO.

As a gamer who grew up with graphical games, I found LambdaMOO’s textuality eerily jarring. Every action had to be explicitly typed out and specified. Even looking at someone was a highly textual and specific action. In the real world and even in graphical games, looking at someone is a largely unconscious act, but in a MUD/MOO, looking at someone was a very deliberate and conscious action. To interact with anything and anyone in a MUD/MOO required an extraordinary level of articulateness and deliberateness. You can’t just nod your head unconsciously. You have to “/nod” deliberately or “/em scratches head in utter confusion”.

I was getting myself totally lost in the hallways and rooms of LambdaMOO when suddenly a female persona starting chatting me up in an isolated corridor. I told her I was a noob, and idle banter turned into a line of questions about having gone through medical physicals for sports or college entry. And suddenly this idle banter became a very serious question of: “Do you want to role-play one out? I’ll be the doctor and you can be the athlete.” Since I had nowhere else to go in LambdaMOO, I said yes. She then teleported us to a private room that she was able to lock. After all, you can’t just go and do crazy things in the middle of virtual hallways. She then warned me, “If you do anything perverted, I’m gonna teleport away”.

What followed in the next ten minutes was the most sexually charged non-sexual experience I’ve ever had in a virtual world. The difference between physical nudity and virtual nudity is that you don’t have to type everything out explicitly when you take clothes off in the real world. Typing out words about your clothing and your body requires a deliberate, conscious articulation that isn’t necessary in physical reality. In the real world, your arms and legs just do what they’re supposed to do. In the virtual world, you have to tell them exactly what to do. And doing so makes you hyper-aware of your physical body via the necessity of articulating your virtual embodiment.

My first experience with virtual worlds was the opposite of Barlow’s. What struck me about LambdaMOO was how textual embodiment was able to highlight and make my embodiment salient in a way I’d never experienced in either physical reality or graphical worlds.

====

Why So Seriosity?

In the spring of 2004, I began working for a start-up that was exploring the intersection of gaming and corporate work. The start-up was later named Seriosity. In the early exploration phase, part of our effort was focused on finding the magic ingredient of online games that we could then leverage in corporate settings. By that time, a couple of undergraduate gamers had also joined the team. What was strange to us at first, but soon understandable, was how the older members of the team latched onto the 3D element as the magic ingredient. Since it was the most salient aspect of online games, it made sense that 3D avatars and objects would appear to be the magic.

On this point, I think no one has said it better than Randy Farmer in 1996:

3-D isn't an interface paradigm. 3-D isn't a world model. 3-D isn't the missing ingredient. 3-D isn't an inherently better representation for every purpose.

3-D is an attribute, like the color blue.

Any time you read or hear about how great 3-D is and how it's going to change everything about computers and services, substitute the word blue for 3-D.

Moreover, there was an assumption that people and meetings would be cooler if everyone had an avatar. The attempt to leverage this aspect of online games may be misguided because it misses the point of how 3D bodies function in MMOs. Games are all about slowing you down so you don’t reach goals instantaneously. Without the need to work to get to the next step, there would be no game. This is why it takes you longer and longer in the game to get to the next level. This is why you have to walk to places on foot. This is why you have to spend weeks playing the game to accumulate enough gold to buy a mount that only lets you move 60% faster. The virtual body that conforms to physical rules is the perfect constraint in an online game because it makes use of the familiar metaphor of embodiment, borrowing many of the limitations of our physical bodies—having to walk to places, not being able to walk through dungeon walls, and not being able to be at multiple places at the same time. Sure, there is customization and status associated with 3D avatars, but as a game mechanic, avatars function as a prop that slow you down by leveraging the familiar constraints of physical embodiment.

Everyone keeps looking for the killer app for virtual worlds, and the only one we know that works so far is gaming. And perhaps the reason for this is because it’s the only application in which slowing people down is a good thing. If you were using a virtual world for work, why on earth would you want people to walk to places, open virtual file drawers, be blocked by virtual walls, or have to figure out what to put on in the morning?

====

Melting Faces

At Stanford, my graduate career was spent conducting experiments in immersive virtual reality. We were interested in presenting what seemed to be physical reality but then changing the rules in the background. We had subjects interact with an agent who shared 30% of their facial features. Or we had subjects interact with an agent who mimicked their head movements at a 4 second delay. Or we created worlds where other participants would appear to be looking attentively at you even when they were looking elsewhere.

In one particular well-publicized study, we morphed a nationally-representative sample of voting age citizens into either Bush or Kerry right before the 2004 election (at a 35% ratio). We found that people were more likely to vote for the candidate they had been morphed with, enough to have swung the election if everyone were morphed with Kerry. What was even more interesting was that in our sample of about 200 participants, not one detected the self morph.

This line of research intrigued me because it foregrounded the possibility of breaking the rules of interaction in virtual worlds in productive ways. Melting faces could lead to an avatar becoming more persuasive. I was intrigued because it reminded me that virtual worlds were good for doing what was impossible to do in the physical world rather than simply produce replicates of physical reality.

====

Do I Look Fat in These Virtual Jeans?

In a lot of academic research and commercial virtual worlds during the years when Second Life was hyped in the media, everyone seemed to be obsessed with replicating physical reality in virtual reality. It seemed important to create avatars that were as realistic as possible, with video-captured facial expressions and animations. People wanted virtual chairs for their avatars to sit in. People wanted virtual cars that their avatars could drive in. And in Second Life, there was virtual food and virtual Abercrombie and Fitch knock-offs.

I think that if we could go back in time and ask John Barlow what world he would make if he could be anything, do anything, and make anything he wanted in a virtual world, it would look a lot like the world on LSD. In other words, it would look nothing like the physical world. What’s so odd about Second Life to me is that in a world where people can be anyone and do anything they want, that Second Life looks so much like Suburban America, except maybe with even more materialism than in real life (which I didn’t think was possible).

On hindsight though, it does make sense. In a world where beauty is a click of the button away and where that form-fitting pair of designer jeans costs almost nothing, it makes it easy to spend way more time and effort on our virtual appearance and jeans than the ones we have in real life. Thus, in a strange way, the virtual world somehow can make us more focused on the physical and material aspects of the world. Virtual worlds make it easier to play out and satisfy the material needs we have in the physical world.

====

Ending Thoughts

Barlow envisioned that virtual reality would take us away from the world of mundane physicality, but I think exactly the opposite has happened. Virtual worlds, in a variety of ways, have succeeded in reminding us of our physical embodiment and accentuating different aspects of our physical existence. Even in virtual worlds, we walk to places, sit in virtual chairs, and buy fashionable jeans.

Our bodies are so ubiquitous in both physical and virtual reality that they don’t often don’t appear to be objects worth thinking about, but the fact that bodies are ubiquitous in both worlds is in fact deeply interesting. What is it that our bodies do in virtual worlds? Why do we need virtual chairs if our virtual bodies never tire? It may very well be that embodiment carried over into virtual worlds because it is a familiar metaphor for interaction. And yet, if we interact and work in virtual worlds by borrowing a physical metaphor, do we end up limiting and constraining the potentials of being and interacting in virtual worlds? More importantly, what new forms of identity, interaction, and work might take place in virtual worlds if we could let our bodies go?

1) Starting: The player has just started playing the game and everything is new and exciting.
2) Ramping Up: The player has learned the basics and is now busy progressing through the content (whether leveling or crafting). They have a sense of where they want to be and are heading for that goal.
3) Mastery: The player is at the higher-end of the game and is either well-situated in a guild and doing raids, or happily soloing high level quests, or competing in PvP content.
4) Burn Out: The player feels like they’ve done everything they can do in the game, or they are beginning to feel burned out from all the raid and social obligations from their guild. They wonder where all the fun went.
5) Casual / Recovery: The player has figured out a way to play the game without burning out. They may be doing intermittent raids, logging in casually to play with friends, casually leveling alts, etc.

====

When players were asked what stage they were currently in, the data found that the largest portion of players considered themselves to be playing at a Casual stage. What is not clear at the moment (and certainly something to explore in future research) is what proportion of these players at the Casual stage have recovered from the Burn Out stage and what proportion have jumped here from the Ramping Up stage (or any stage before Burn Out). Indeed, while there is a loose assumption of going from one stage to the next, it is likely that some players jump from Ramping Up to Burning Out (among other jumps).

While it may at first be surprising to see so many players classify themselves in the Casual stage, there are reasons this may be the case. From the player narratives in the earlier study, there was a sense of inevitability of burning out once ramping up had begun. Players seemed to inevitably hit a threshold where additional advancement required an effort that became more tedious than fun. Thus, in long-running MMOs like WoW, the majority of players who have been playing for more than 1 or 2 years must therefore have found a way to get off the grind path and still enjoy the game in a more casual play style.

====

While the gender differences were quite small, there was a bit more interesting variation across age groups. In the chart below, we see that younger players are most likely to be in the Burn Out stage. This is consistent with data we’ve seen elsewhere that younger players are more achievement-oriented and thus more likely to follow the grind path into the Burn Out stage. Overall though, it appears that gender and age differences were quit minimal.

====

I also asked players how long they had been currently playing their MMO in months. Below is a stacked chart that shows proportion of each stage across months played. One thing to keep in mind when reading this graph is that all the players who have quit are not part of these stacks. Thus, these aren’t numbers about all players, but rather, the players who have not quit the game by the 5th or 7th or 12th month.

Until the 3rd month, most players are still in the Starting and Ramping Up phases (68% combined). By the 4th month, the majority of players are past the beginning stages (57% in Mastery + Burn Out + Casual). By the 7th month, no one considers themselves to be a total noobie in the game and over one-third of players consider themselves to be playing casually. At the one-year mark, we see the largest proportion of players at the Mastery stage. And by the 19th month mark, the majority of players (54%) are in the Casual stage.

====

Motivation scores across the Achievement, Social, and Immersion factors were also collected from players, so we can take a look at how motivations for playing change over time. First below is the chart of Achievement scores. Throughout the stages, men tend to be more achievement-oriented than women, but for both genders, achievement is most compelling during the Mastery and Burn Out stages. What’s interesting about Achievement as a play motivation is that it may directly lead a player to the Burn Out stage (thus both a good and bad thing at the same time). As the chart also shows, players who make it to the Casual stage have let go of their strong Achievement drives. Or in other words, strong Achievement players in Burn Out stages tend to quit instead of stay, and thus the ones who remain in the game (in Casual stage) score lower on Achievement on average.

We see a somewhat similar pattern with the Social scores with some interesting differences. Both men and women actually start out not caring much about socializing. It is only starting from the Mastery stage that both genders care a more about social stuff in the game. Also at this point, there is suddenly a large gender difference in the Social motivation. In contrast with the Achievement motivation, the Social motivation remains somewhat important among players in the Casual stage. This is consistent with the saying that “People start for the game, but they stay for other players”.

====

What’s interesting about the Immersion scores are they are the exact opposite of what we’ve seen so far in the general pattern of Achievement and Social scores. Players start by caring a lot about being immersed in the game, but by the Mastery and Burn Out stages, they care a lot less about it, and this rebounds a bit in the Casual stage. We also see an interesting gender difference throughout. Women have more extreme feelings about Immersion than men. They like it more than men in the initial stages, but they also dislike it more during the Mastery and Burn Out stages.

Looking back over these three motivation factors, another interesting observation can be made. Both the Achievement and Immersion scores show a similar rebound curve. What is important during the initial and ending stages are not important during the middle stages. On the other hand, for the Social motivation, what becomes important in the middle stages remains important in the end stage.

While this set of data provides some information on player life-cycles, it also raises many other questions that are still unanswered. For example, as I noted early on, it’s not clear how players get to the Casual stage. Is it truly an end stage after recovering from Burn Out? Or a stage that players can jump to from any other stage? There is also an issue of whether players at the Casual stage may get sucked back into the Mastery stage, and how the cyclical nature of the stages may play out. From a data perspective, the constant attrition of players over time may also distort our understanding of what is really happening by only looking at the players who stay. For example, do players in the Casual stage have low Achievement scores because they have actually changed or because strong Achievers have all quit the game by then?

Below is how the data came out by gender. Overall, it is what is typically referred to as “casual content” that is most highly rated (i.e., solo content and small raids). Note that in the survey, “small raids” was defined as regular 5-man dungeon instances. Another observation is that where there were some gender differences, the general order is roughly the same for both genders. Rated the lowest were high-stress PvP content—duels and open PvP. Content types that require a high level of management and coordination (i.e., large raids and large-scale PvP) received average ratings.

====

Looking at the data by age reveals other interesting findings. Below is a chart that plots the correlations between each the content type ratings and age. Note that it is the correlation coefficient that is plotted. Thus, a “0.20” means that age correlates with the appeal of the content type positively. The graph below shows that older players are more likely to prefer solo content, while younger players are more likely to prefer high-stakes PvP and large-scale raids and PvP. We’ve seen this “hard-core” aspect of younger players in other places before.

Of course, despite this underlying granularity of content types, most MMOs still ask players to choose from one of about four server types: normal PvE, PvP, RP PvE, and RP PvP. Here’s a chart that shows server choice by gender from the survey.

====

We can see several patterns in this chart. First off, normal PvE servers are the most popular server type (despite the fact that these are “role-playing” games). Secondly, PvP servers (whether normal or RP) are twice as popular among men than women.

And here is a stacked chart that shows server choice by age group. This chart shows again the age difference in PvP appeal we’ve seen above. The two RP server choices (the lower two colors) actually remain mostly flat across the age range and it is the PvE vs. PvP data that shifts. Appeal of PvP servers peaks in the 18-22 age group and mostly falls after that point.

What this set of data reminds us of is that server differences aren’t only a function of server rules, but that they are also a function of demographic differences. The difference between a normal PvP and a PvE server isn’t just because of the rules, but also partly because PvP servers attract significantly more men relative to women and it also draws from a younger age group.

- Post a message on your community forum or message board about the findings.
- Create a link to "The Daedalus Project": http://www.nickyee.com/daedalus/
- Add a link to The Daedalus Project in your forum signature.

As usual, your comments and feedback are most welcome. Any questions should be directed to contact@nickyee.com

Let’s start with the basics. 89% of respondents said that at least one of their characters belonged to a guild. There were no age or gender differences in this percentage. 25% of these players belonged to more than one guild. Female players were more likely than male players to belong to more than one guild (32% vs. 22%). There was no age difference in this percentage. So the majority of players who belong to a guild only belong to one guild.

Thus, it’s not surprising that the majority of respondents (87%) were able to identify one guild that they had a strong affiliation with. For the remaining data, respondents who could identify a strongly affiliated guild were asked to use that guild as the reference. For the respondents who could not, they were asked to use the most recent guild they joined as the reference.

Players were most likely (23%) to be in guilds that they themselves created (either alone or together with other players). Here are brief explanations of the different categories:

RL Friend: Someone the player knew in RL invited the player into their guild.
Asked to Join: The player saw an ad in general chat or an online forum and asked to join the guild.
Online Friend: Someone the player knew from online invited the player into their guild.
Invited after Interaction: The player was invited to join after grouping with, chatting with, or interacting with a member of the guild.
Random Invite: The player was asked to join the guild with no previous interactions with members of the guild.
Split / Merger: The player was in a previous guild that splintered into this current guild or was in an old guild that merged with another guild to create the current guild.

====

In terms of guild migration across games, 26% said that their guild existed (or currently exists) in a different game. Of these respondents, 56% had been part of the guild in a different game and migrated over, while the remaining 44% had joined only in the current game. The overall remaining players (74%) noted that their guild did not exist previously in any other game.

Most players have been in their guilds for 1 year or less (64%), although about a quarter of respondents have been in their guilds for more than 2 years.

When asked to estimate the number of active players in their guilds (excluding known alts), the average guild size was estimated to be 38.9. The median was 30. And about a quarter of players had 10 or fewer players in their guilds. This is significantly higher than the number we’ve seen in the PARC PlayOn data. It’s not clear though how much this is due to selection bias and how much it is due to shifting guild composition.

====

The majority of players have someone in their guild that they know in RL, whether this is a friend, a colleague, a family member, or a romantic partner. The most common categories were friends (59%) and romantic partners (35%).

In fact, the majority of players (59%) joined guilds where there was someone they knew in RL (i.e., before they joined the guild). This also lends support to data we’ve seen elsewhere that MMOs play a role in maintaining RL social connections.

====

In terms of the general tone of their guilds, players tended to describe them as being more on the casual side. Overall 52% rated their guild as being casual, 6% as being serious, and 42% as being somewhere in-between. There was no gender difference, but older players were a little more likely to describe their guilds as being casual.

To get a sense of how raid-oriented their guilds were, I asked respondents to categorize their guilds in one of four tiers. About half of respondents were in guilds that are too casual to raid or just started to learn how to raid. The other half considered themselves to belong to mid-tier or high-end raiding guilds. There were no age or gender differences in this distribution.

====

And finally, I asked respondents how happy they were with their guilds. Unsurprisingly, most players were very happy or extremely happy with their guilds (which they had previously noted they had a strong affiliation with). Female players were much more likely to be extremely happy with their guild than male players. This is consistent with trends we’ve seen elsewhere where female players tend to be happier with the game in general and less likely to quit in the near term.

How do people decide what character to create? Do they always pick certain character classes? Do they carefully read through class descriptions before deciding? Do they go by their gut reaction? To get players to talk about the process they use, I asked respondents to imagine that they had just joined a new MMO and were about to start a new character. I asked them to describe the process they used to create their first character and in particular what factors were most important to them in this process. We then took the open-ended responses and came up with a set of codes and then coded through 500 responses. Below is the distribution of codes, in descending order, along with examples of what players said.

Class Type (20%) – Players in this category have a certain class type they prefer and will always go for that type when it is available.

Class Abilities (17%) – Players in this category look for particular aspects of character classes that are often independent of class types. This may be the most overpowered class, the class that is most challenging to play, or the class with the most crowd control abilities. In a sense, these players are trying to maximize for a factor that is sometimes but not necessarily tied to particular classes.

Race Aesthetic (11%) – Players in this category look through the available races and pick the race that most interests them as their first decision.

====

Game-Play Factors (10%) – Players in this category are trying to optimize for a game-play factor, whether it is soloability, utility in groups, end-game utility, etc.

Class Research (7%) – Players in this category have no preference for class types or class abilities, but they base their character creation primarily on choosing a class. So they first read through all the class descriptions and then pick the one that most interests them.

General Appearance (5%) – Players in this category use character appearance as the primary decision factor in creating their character. They try to find a race/class combination that has a look that appeals to them regardless of functionality.

Specific Combo (5%) – Players in this category have very specific race + gender + class combinations that they will go for first if available. Oftentimes, players in this category are role-playing and have a character concept in mind that they are trying to fill.

====

The remaining categories have fewer than 5% of codes and will be described briefly without examples.

Race Type (4%) – Players in this category always go for a particular race if it is available.

Companion Complement (4%) – Players in this category create a character that best complements a friend or romantic partner they are playing the game with.

Gut Reaction (3%) – Players in this category do what feels right and fun at the moment.

Uncommon (2%) – Players in this category research what the most uncommon character is and create that.

Alts (2%) – Players in this category will create many characters at once to try them out and then after a while pick one as their main.

Racial Abilities (1%) – Players in this category will first look at racial abilities and start there.

Starting Area (1%) – Players in this category will find the starting area that appeals to them most and go from there.

Given the large number of codes and the limited number of coded responses, I did not look into age or gender differences as the cell sizes would have been too small to be reliable.

500 of these responses were then coded. The main difficulty that emerged in coding these responses was that players described their preferences at different levels of abstraction. For some players, they only cared that it was a class that could cast spells. For others, only that it was a mage class. And for others, they really only wanted high DPS glass cannons. To give the best sense of what players said, I’ll show here the codes at the specific level of abstraction that players used in their responses and try to group them in as coherent a way as possible (although hybrids make this difficult).

Given the large number of codes and the limited number of coded responses, I did not look into age or gender differences as the cell sizes would have been too small to be reliable.

Note: Be careful of interpreting this table as saying that healers are most popular. It's more that people conceptualize healers with fewer levels of abstraction. The rows in this table are at different levels of abstraction. In other words, the people who chose "Melee", "Tank", "DPS Melee", "Paladin", and "Melee Hybrid" would probably have all chosen the Warrior class if forced to choose one of the four classic RPG archtypes.

On average, players have 8.7 characters on their accounts. As the graph below shows, female players tend to have more characters on their accounts than male players (9.7 vs. 8.3). There does not appear to be a stable age trend.

While this number of characters may suggest that it would be hard for players to have to pick a main character, most players have no problem doing so. Overall, 89% of respondents said that they could readily identify a main character. There was no gender difference, but younger players found this easier to do than older players (96% vs. 85%). This may be linked to a higher achievement motivation among younger players that makes it easier for them to focus on a few characters. We’ll see more data later on that supports this.

====

In terms of level spread, the majority of respondents (70%) said that they had one character that was significantly higher in level than their other characters. There were no age or gender differences in this response. And overall, 68% of respondents had at least one max level character. There was a mild age trend where players in the 23-28 age group were most likely to have a max level character and those at the two ends were less likely to do so. Surprisingly, female players were also more likely then male players to have at least one max level character across all age groups.

In terms of time distribution among characters, most players played more than one character in the past month. Only 17% of players had only one active character in the past month. On the other hand, 76% of respondents noted that they have a character that they devoted over 75% of their playing time toward in the past month.

====

When asked more directly whether they tend to devote their playing time to one character or split it among many, only a handful of players (12%) said that this was their primary play pattern. The greater part of respondents noted that they were somewhere in between. About 40% of respondents noted that they tend to devote their playing time towards only one character. Male players were more likely than female players to devote their time to one character.

If we slice this data by age, we also see that it is younger players who tend to devote their playing time towards one character. This may be due to their tendency to find achievement motivations appealing (thus having a stronger drive to focus their energy in one character to level them up high) and this helps explain the earlier finding where we saw that younger players also found it easier to identify a main character.

====

To identify which of these factors players used to identify their main character, I used a classification tree analysis (a standard machine learning technique). The analysis showed that the logic sequence is best capture by the following branching:

- Is there a character that you spend 75% or more of your playing time towards?

- If yes, then that character is the main character.
- If no, then is one of your characters significantly higher level than all other characters?

- If yes, then that character is the main character.
- If no, then the player can’t identify a main character.

The data show that asking players to identify a main character is something almost all players can do, but that players may use different rules to pick their main character. At the same time, the data show that asking players to pick their highest level character or the character they spend the most time on specifically may result in cases where the character isn’t the one that the player would have identified as the main. In a world where every player has multiple identities, the concept of “you” can get very tricky.

I started with a question about game genre. I asked players to rate a list of possible genres. It wasn’t too surprising that the Tolkien genre came out top as most MMOs currently are in that genre. The only significant gender trend was in the futuristic spaceship genre that was preferred by men. Of course, one thing we have to keep in mind with this and other charts below is that oftentimes it may be hard for players to imagine comparable games in genres that aren’t well represented.

Pirates of the Burning Sea recently came out and I was curious how players would distribute themselves among the factions. Even though about half of the respondents preferred to be pirates, it’s good to know that about the same percentage of players would rather not be pirates (and thus the pirates vs. non-pirates factions would be evenly matched at least).

====

In a three-part division of magic, nature, and technology, it was interesting to see almost an even split among the three. Women more strongly preferred the nature tribe while men more strongly preferred the technology tribe. The magic tribe came out even by gender.

In a hypothetical game where you could either be the vampire or the vampire hunter, the majority of players chose to be vampires, but women were more interested in being vampires while men were more interested in being the vampire hunters.

In a game that allowed a warrior role, a merchant role, and a spy role, respondents preferred to be spies, warriors, and merchants, in that order. Men were about twice as interested as being warriors while women were more interested in the non-direct combat roles.

====

In the classic RPG class types, we see a fairly even distribution. The stereotypical gender difference is also seen. Men prefer to be warriors while women prefer to be healers. There were no gender differences in the archer or mage classes.

In many games where there are warring factions, disparities between the faction populations typically arise. Given the choice, it was a little surprising to see that 80% of respondents would prefer the minority.

And finally, here’s a classic trend we’ve seen before. When asked whether they would gender-bend, men were about 4 times as likely as women to create a character of the opposite gender.

Dmitri Williams led a project where we analyzed server-side data from EverQuest 2 in combination with a survey of randomly sampled players from the game. Our first overview paper of the findings just came out in a peer-reviewed journal and it is accessible to the public. While the BMI (body mass index) finding has been making the rounds in media, that finding is just one of a large set of findings from the paper. For example, what’s surprised me the most is how closely my self-selected surveys came to the findings from a randomly sampled survey in terms of average age, hours played per week, etc. So even though there clearly is bias in the self selected surveys, it appears that the findings track fairly closely to more rigorously sampled data.

In my work with the Incredible Internet, we’re starting a series of three video podcasts for parents in helping them understand online games and how to set reasonable expectations for their kids. The first podcast just came out. My main goal in helping create these videos is to strike a balanced tone—describing the main risks and suggesting sensible rules without resorting to loaded words and fear-based messages.

And a little tangential to gaming, but relevant to virtual worlds and the upcoming presidential election, the local ABC channel did a clip on our work at the Stanford VR lab highlighting the presidential face morphing studies we’ve done. That clip is also available online.