The chapter is on how the introduction of a threat meter addon changed my raid group’s practice over time.

Continued from http://markdangerchen.net/2010/01/19/enrollment-of-threat-meter-addon-work-in-progress/ and http://markdangerchen.net/2010/04/06/enrollment-of-threat-meter-addon-part-2/

About four months into our raid’s life, in February [or March?] of 2006, we started using a new addon called “KLHTM” or “KTM.”

Created by a player named Kenco, KTM did the work of keeping track of which abilities a particular player used while fighting a monster, how much threat those abilities generated, and then visually displayed that information to that player. What’s more, any instance of KTM could talk to other instances of KTM installed on other people’s machines and thereby aggregate all of the threat data for all players who had the addon installed, displaying relational charts of everyone’s threat level to each player.

<more after break>

[image of KTM here]

This allowed the offloading of human cognition to a nonhuman resource, effectively eliminating much of the guess work that went into World of Warcraft fights.

Before the addon, my raid group had progressed to the last boss in Molten Core. The write-up about our practice found in the Communication, Coordination, and Camaraderie paper describes how our chat was multi-threaded and interleaved, hierarchical and specialized, roughly divided by class role. Among many other things, one thing this allowed us to do was to be highly coordinated in our tactical take-down of a raid boss. By the time KTM was introduced, we had become quite proficient in dividing up our attentional resources and communicating along certain channels, escalating which channels were in use when necessary. After KTM became the standard, the necessity of using those chat channels was not as acute as before. Suddenly, any player of any class could keep track of the threat generated of all the other players. Not only did the addon help us with our cognition, it’s use also forever changed who communicated with whom about what, most notably allowing raid leaders to caution specific raiders about their threat generation. This effectively substituted knowledge-based trust in others with a technological advancement where trust or faith in other players’ ability to manage their threat didn’t matter. Yet, at the same time, KTM let us be much more efficient in our monster killing. We could ride the edge much more effectively, thereby taking down monsters faster than we had been before, which also lowered the learning curve associated with new encounters.

Kenco’s Threat Meter is an interesting example of Latour’s recognition that objects within an activity system may have multi-layered complex histories. The emergent network or arrangement of the objects in circulation, likewise, is complex and multi-layered, both in a micro to macro scale of physical closeness and across multiple timescales. [find quotes from Reassembling the Social to use here.. multiple ways of thinking about the shape of the network] KTM’s history is rooted in a gaming tradition of deconstructing [decrypting/decoding?] the underlying mechanics or math of a game, which, as a practice, probably existed shortly after the first game. Games, after all, essentially present players with some sort of system of rules or simulation to uncover. Pattern recognition is the main learning activity a gamer does. Early widespread understanding and taking advantage of the game rules probably came about with the rise in table-top role-playing games, most notably Dungeons & Dragons, where the practice of creating a character that exploited the game mechanics was called min-maxing–minimizing resources spent on relatively useless abilities and skills to maximize resources spent on the most effective abilities and skills. This was only possible after a player was able to grasp the underlying mechanics and figure out particularly effective combinations of abilities for specific situations. With the rise of digital role-playing and strategy games (particularly Starcraft) and access to web forums where players could discuss, debate, and co-construct their models about various game mechanics, the practice became known as theorycrafting, taking the name from the IC [double check this and reference Chris Paul's work maybe].

[something about how D&D became about numbers rather than role-playing?]

Kenco was one of the early theorycrafters for World of Warcraft. In January 2006, he posted to the WoW European web forums that he thought it was possible to run a number of in-game tests, systematically testing out different variables, to uncover how WoW calculates threat. At the time of his posting, in fact, he had run several of these simulations, and he proceeded to discuss his findings, dispelling quite a few myths about threat generation. This was counter to the general thought that exact threat mechanics were forever going to be hidden from the player community.

It’s often said that we will never be able to work out the way threat and hate lists and mobs’ AI works, because it’s too complicated and unknowable, that we’ll only ever have crude approximations and guesses. I’ve conducted some decent, rigorous tests, and i have what i believe is a good list of hate values and explanations of gaining and losing aggro and the behaviour of taunt. I am also able to debunk a few myths about how threat works.

http://www.wowwiki.com/Kenco’s_research_on_threat

After carefully describing his major findings, he gave a list of suggestions for strategies to use in future fights and then ended his post with this: “There’s no amazing super secret randomised blizzard aggro algorithm. The concepts are simple and the values can be fitted with nice numbers. Even formulas for threat-reducing knockbacks can conceivably be worked out, if threat values are carefully monitored.”

In February [or March?] players started testing out Kenco’s first stabs at a threat meter addon, and on March 1, 2006 (according to Curse’s records), he released the first public version of KTM to Curse.com, a website devoted to hosting a World of Warcraft addon repository.

Since then, theorycrafting became common practice, probably most popularized by the site ElitistJerks.com, where class-based discussion boards devoted to damage and threat calculations feature players using sophisticated spreadsheets and custom tools to model and number-crunch every known, manipulable in-game variable. Figuring out threat and then exposing the underlying model to all players via the addon became so successful and so widely adopted into common raiding practice that Blizzard began to tighten up their raid encounters to depend even more on players’ ability to manage their threat and aggro levels.

[something about how WoW became about numbers and not role-playing... parallel to D&D's evolution]

The chapter is on how the introduction of a threat meter addon changed my raid group’s practice over time.

Continued from http://markdangerchen.net/2010/01/19/enrollment-of-threat-meter-addon-work-in-progress/

[Need an illustrative, hypothetical table here?]

Looking at Rogues in particular, since I know the game best from their point of view, having played a Rogue during my time with the raid group, I can say that we did not know exactly how much threat each of our abilities generated, but the Rogues did know that certain abilities generated much more threat than others. These were roughly correlated to the damage output of the various abilities. For example, we knew that our main attack, Sinister Strike (SS), generated a consistent, predictable amount of threat that was safe to use, whereas, Eviscerate generated much more threat since generally its damage output was much higher. Yet, the use of Eviscerate was balanced with the fact that we could not use it as often as SS.

<more after break>

Rogues operate on a mechanic of building up or chaining “main” attacks that enable the activation of what are known as “finishing” moves. Sinister Strike is one of these main attacks that can be activated in a sort of rhythmic fashion every three seconds or so [double-check timing of abilities], building up a “combo point” with each successful hit. Rogues can build up to five combo points with these main attacks. Eviscerate is a finishing move that spends or uses up the built-up combo points, and it does more damage with more combo points, giving Rogues an incentive to build up five combo points before using Eviscerate. Thus, Eviscerate is generally used less often than SS, in a more syncopated rhythm, but when it is does get activated, it does more damage.

If we were to graph the damage output of a Rogue using SS and Eviscerate over time, we’d see a baseline level of damage from SS and spikes in the graph every twenty seconds or so from Eviscerate. Since we’ve correlated damage output to threat generation, our threat graph follows a similar pattern with a baseline, consistent threat level that includes spikes in the threat generation activity every twenty seconds or so.

[Insert hypothetical graph here]

These spikes in threat generation were known as danger zones where we needed to be cautious and alert in case the mob aggroed on us. Well, we would have, except for the fact that it was general consensus that for certain fights, especially with boss mobs, we shouldn’t use Eviscerate at all. Instead we used Slice and Dice (SnD), a different finishing move that did not output damage in spike form. Rather, SnD made our non-activated attacks faster.

Every character has a default attack that doesn’t require any input from the player. The level of damage from this default or “white damage” (so called because it is displayed in white in the in-game combat logs) attack from Rogues is determined by the speed of how often a Rogue swings his or her weapons, which is determined by the speed factor or attribute of each weapon [are Rogue's default speed faster? double-check this], multiplied by how much damage the particular weapons could do with each hit. The resulting number is known as the weapons’ damage per second or DPS, a term that has been, as has been mentioned above, co-opted as the name of the role Rogues and other damage dealing classes assume. So, the baseline in the graph above is actually a combination of the white damage plus the consistent damage from SS (a form of “yellow damage,” the color of damage coming from activated abilities in the combat logs).

Slice and Dice temporarily speeds up a Rogue’s default attack frequency, thereby raising the baseline damage by increasing white damage without adding spike yellow damage to the graph.

So, for many boss fights, the Rogues would generally stay away from using Eviscerate and instead use Slice and Dice because we did not want to have spikey damage graphs for fear of having spikey threat graphs, assuming that spikes in threat generation were more likely to pull aggro away from our tanks due to their less predictable nature.

But, again, all this was sort of kept in our heads, and, as a rule, using SnD was not strictly adhered to by all Rogue players. This is especially true while we were learning new boss fights. Often, in order to succeed, we had to push the limits and continuously ride on the edge of too much damage / threat. If we weren’t on the edge of our ability, like an Olympic skier, then we were under performing, which lead to a possible raid wipe if the raid healers were going to run out of mana trying to maintain our current practice. Yet, like all the Olympic skiers who wipe out, which happens quite frequently, we were always in danger of going over the edge or pushing too hard.

The first few times we encountered a new fight, raid wipes were expected. This was to allow for us to learn what mechanics were involved with the new monsters but also to allow for us to test the limits how much damage or threat we could generate. In danger of using a racing reference too heavily, it’s like we’re learning the course or track for the first time and need to adjust our speed if we discover that our first attempts were too fast or too slow.

This is not to say that failure was always welcome, though. Even though early wipes were seen as learning opportunities, it was frustrating to wipe over and over again in the same game session.

Anyway, all this leads up to our fight with the last boss in Molten Core, Ragnaros. When we first encountered him, it was generally agreed upon by the Rogues in the raid that we should stick with using Slice and Dice to maintain a consistent, predictable level of threat.

[quote from Roger here?]

But as we were learning the fight, something completely new changed raiding in World of Warcraft forever.

[Note: After writing, I've realized that I don't really go into much detail, so I probably could've written this stuff down a long time ago, but I wasn't sure what I'd write, so...]

There were two cases that I’ve previously written about or mentioned. First, there’s the two cases of inclusion that I wrote about in “Ethical tensions between the roles I play” where guildies weren’t quite fitting into the social norms of the guild. Second, there’s the case of a guildie who was gkicked due to an argument over loot rules. I wrote about this in “Play my way,” the chapter co-authored with Lisa Galarneau that was originally to appear in a book on the politics of play in virtual worlds. That book never got off the ground, so we recently revised it and submitted it to the Handbook of Research on Improving Learning and Motivation through Educational Games.

Anyway, one case I haven’t really shared before was when a woman in my guild claimed she was going to attempt to commit suicide. An interesting thing to note about this case was that it occurred during the atypical night of raiding that I wrote about in the “Communication, coordination, and camaraderie in WoW” paper in Games and Culture. The timing of her call-out for help significantly contributed to me not paying as much attention to the raid as I normally did, which, in part, added to the “off night” feeling that led to our nigh raid meltdown.

<enter description of suicide attempt here> [Note: maybe some day I'll write about it fully, but I've realized that I don't need to describe it more now...]

All of the above events were ones that I wasn’t expecting to encounter while managing a guild.

There’s a nod-with-a-smile-knowing-better-now-that-I’m-wiser kind of feeling when I think back at the time of our guild formation. There were five of us who were friends outside of the game. The only reason I was the guild master was because I took the time to run over to the guild house and buy the charter. Other than that, my thought was always that us five were a quintumvirate or pentumvirate or whatever word is used with a ruling body composed of five people. When we were naming the guild ranks, I chose Overseer for guild master, mostly trying to stay in character as we were Horde-side on an RP server (with the other ranks being Officer, Veteran, Grunt, and Peon). Yet, the nominal marking of me as the Overseer positioned me in a role that came with additional responsibilities. These were projected onto me from both guild officers (the other four members of the ruling body) and the regular guildies. Often when tension would occur, a couple of the other officers would argue (somewhat jokingly but somewhat seriously) that I should handle the situation because I was the Overseer. Eventually, I came to understand that it was easier for guildies to see one person as the de facto leader, and I began to accept the de jury role as my actual role.

T.L. Taylor suggested I expand on my “Ethical tensions” paper to focus more on the mediating role I found myself in, between officers and regular guildies. Not only did I have to deal with the problem guild members, I also had to figure out a way to reconcile competing opinions among the officers about how to do that management work. And, of course, I had to do this partly because of my title but also partly because I felt obligated to intervene on behalf of the guild members such that they were being treated fairly and responsibly. This compulsion to ethical behavior was in tension with what seemed like the norm (or stereotyped norm) of game group management, which was to just boot the non-socializing guild members and be done with it (AKA just boot the fuckers).

Our guild prided itself on being relatively inclusive and flat, though one officer put it nicely when he pointed out that the emphasis was on the word “relatively,” since we *did* in fact exclude those who clearly did not fit in–mostly people who weren’t articulating and communicating effectively. To be and feel included necessitated a certain level of communication and social awareness. So, when it was suggested by other officers, to just gkick the problem guildies did not seem to me to be in line with our guild credo. It was one thing to not invite someone to be a member of the guild because it was clear he or she didn’t fit in, but to kick someone out once he or she actually was already in needed justification. It needed to be explained and described with specific examples of problem behavior so the non-socializing guildie could present a counterargument.

But those values of negotiation were in full conflict with our notion of what game playing was. We were there to have fun–not to work, not to deal with drama. Again, I can smirk now, knowing what I know and having a better sense of what it means to play in an online game. Play *is* work. Playing with others necessitates negotiations of roles, responsibilities, and social norms. With enough others, there’s going to be conflict or misalignment. That’s a given, and the world won’t get better by refusing to deal with conflict. Refusal to resolve conflict excludes those who need a help-up in participating legitimately in the community. (Are you down with LPP? Yeah, you know me…)

One could argue that it all comes out in the wash, that people who are excluded from one group eventually find another where they fit in. That’s the beauty of a critical mass of people in an online space. But I think this possibly leads to the forming of more and more insular groups, and I consider it a problem that perpetuates the intolerance we have in offline life. That’s not the world I want to play in.

Nick Yee once asked me why I wanted to help the problem guildies become acclimated to the guild and/or find an alternative suitable guild for them to play with. When he asked me this, all I could say was that I felt compelled to help others when I could. Maybe I thought that as someone who valued education, it was hypocritical not to want to help others. (Channeling Malaby) I can now say that acting in social settings is always a contingent move towards the display of cultural capital. And as a person who values diversity, inclusion, and democracy, I sometimes have a strong compulsion to help others learn how to be successful in their contingent acts of play/work. This was especially true when I was positioned within my guild as the Overseer who had accepted certain responsibilities to live up to the guild’s stated values.

(This all sort of speaks to positioning theory… I was positioned by others into the role of someone who is in charge of handling guild conflicts. I (re)positioned myself into a role that compelled me to mediate conflicts in a way that I thought was ethical and fair. Who was doing the positioning? Did I accept or rebel or simply transform my position?)

In case you’re interested in academic references and/or don’t feel like clicking on the in-line links:

• Chen, M. (2009). Communication, coordination, and camaraderie in World of Warcraft. Games and Culture, 4(1), 47-73. http://dx.doi.org/10.1177/1555412008325478
• Galarneau, L. & Chen, M. (in review). Play my way: The politics of cooperation in massively multiplayer online games.
• Harre, R., Moghaddam, F. M., Carnie, T. P., Rothbart, D., & Sabat, S. R. (2009). Recent advances in positioning theory. Theory & Psychology, 19(1), 5-31. http://dx.doi.org/10.1177/0959354308101417
• Holland, D. & Leander, K. (2004). Ethnographic studies of positioning and subjectivity: Narcotraffickers, Taiwanese brides, angry loggers, school troublemakers. Ethos, 32(2), 127-139. http://doi.wiley.com/10.1525/eth.2004.32.2.127
• Lave, J. & Wenger, E. (1991). Situated learning: Legitimate peripheral participation. Cambridge University Press.
• Malaby, T. (2009). Making virtual worlds: Linden Lab and Second Life. Cornell University Press.

The Enrollment of a New Actor and the Redistribution of Responsibilities in a World of Warcraft Raid Group

In World of Warcraft, each individual actor in a raid group is in charge of certain tasks and responsibilities. At one point in the life of the raid group I studied, a new actor was allowed into the group. This newbie rendered new services to the rest of the group. The services rendered were essentially rating the actions of the others in the group—that is, assigning a specified number value to their actions—and then remembering who did what to add up the ratings from each particular player. This newbie, though, didn’t actually care one way or the other if these services were used by the others, but if another decided to use them and have his or her rating displayed, that player had to abide by new rules associated with these new services. The newbie wouldn’t verbally announce others’ rating. Instead, a sign was held up and players had to manually look over to read what their ratings were. In that way, the newbie not only served but also demanded, not only taking on the burdens assigned with this new role but also prescribing new responsibilities on the others. Yet others in the raid group, first slowly then readily, came to adopt the use of these new services into their practice as the services’ benefits became increasingly clear. The group came to consider the new tasks as essential parts of its raiding activity, and players could barely remember a time when the rating-remembering services were not used. The newbie became one of them—not a newbie but a veteran—and the group merrily went on its way. But this veteran wasn’t one of them. In fact, it wasn’t even human. It was a technological device, a program, a construct, an “addon” modification to the game.

(More after the break.)

This paper documents the enrollment of this nonhuman actor and its history within the raid group that I studied. The addon was instrumental in helping the raid group become efficient and successful with many in-game battles. Interestingly, the addon played only a temporary role in the raid group’s assessment of a specific encounter, the last monster, Ragnaros, in a fiery cave system known as Molten Core. It helped the group by testing and ruling out a possible diagnosis of the problems with the group’s strategy. After eliminating that possibility, its use was no longer necessary, since its original intended role never needed to be filled in the fight against Ragnaros.

This paper helps us see that, within a learning space or network, people and their material resources collectively share responsibilities, and the distribution of these roles and responsibilities change over time as new challenges are met and as new actors enter the network. This is a story, in other words, of how a network is disrupted by unexpected events and the redistribution work done by the network’s dynamic, adaptable actors to overcome those events.

Roles, responsibilities, and aggro

Each character in WoW fits into an archetypal role based off of historical precedent in the fantasy role-playing game and MMOG genres. In representation, characters are warriors, mages, priests, etc., but for the purposes of the underlying game mechanics, these various hero classes can be roughly categorized into a function-based triumvirate consisting of “tank,” “healer,” and “DPS.” Each of these categories has specific duties and responsibilities to carry in a raid battle. Tanks, with their plentiful health points and massive armor, must keep the monsters occupied and focused on them while healers continually spend mana or magic points, casting spells to make sure the tanks stay alive. DPS (shorthand for damage per second, a way of valuing damage dealers) can then go about actually killing the monsters.

Each category of roles in the triumvirate is therefore necessary to be filled for a raid group to be successful. Without tanks, the healers cannot possibly cast spells fast enough to keep whoever is being attacked alive, and the monsters will kill everyone rather quickly. Without healers, the tanks will die, and the monsters will, again, chain-kill everyone. Without DPS, the healers will eventually run out of mana, the tanks will die, and the monsters will ultimately kill everyone.

The problem is that a monster generally attacks whomever it deems is the most threatening to their survival. If a DPS hits a monster particularly hard or a healer heals too effectively, the monster may take notice and decide to hit back. Whoever has the monster’s attention is said to have “aggro,” and the monster switches targets when players “steal aggro” from others. Tanks can try to prevent this by activating various abilities meant to maintain aggro, while the DPS and healers try to keep their performance at an even, consistent, predictable level without spikes that would make the monster take notice. In other words, many of the encounters in WoW, and indeed most MMOGs, are a balancing game where the three roles of the triumvirate work to maximize their efficiency while keeping the tanks the focus of the monsters attention. The fights, therefore, are engineered by the game developers to test and destabilize the triumvirate.

So, each role in the triumvirate (tank, healer, DPS) has specific responsibilities in a fight. Yet healers and DPS cannot “go nuts” with their abilities, “spamming” their most powerful ability over and over again. Rather, they are constrained by the need to make sure the tanks maintained aggro.

Threat management

These games must obey some sort of algorithm, and, in this case, the way in which a monster decides who to attack is completely reactionary to the actions of the raid members. The underlying “brain” of the game creates a table that includes a row for each raid member, and in each row is a number that starts off at zero and increases a certain amount every time that particular raider activates an ability. The amount depends on the ability. This number is called the threat level. One of the jobs of the raiders, then, is to make sure that the tank(s)’ threat level is higher than everyone else’s.

When the raid group I was part of first started, we each had to internalize our threat level and play it by ear, so to speak. There was no common resource or explicit knowledge of specific numbers associated with specific abilities. We knew from experience that some abilities generated more threat than others, and we had to weigh their costs against the benefits of the abilities. Very often, when a player died, it was because he or she stole aggro from the tank(s). That is, he or she misjudged how much threat was being generated and accidentally raised his or her threat to a higher level than the tank(s)’ threat level. If this happened enough times during an encounter, it usually ended up as a raid “wipe,” where everyone in the raid group died.

It’s an odd piece, but I’ll write about it later…  right now I gotta go hang out with SG. :)

A flurry of activity occurred commenting about the use of ANT. It’s not a method but a framework, for example.

I was excited because I am thinking of using ANT to look at WoW raiding practice, and since I wanted to get feedback, too, I posted the following:

Hey all!

Fascinating discussion.

I’ve recently starting reading about ANT and have been toying with the idea of analyzing how a raid in WoW works through an ANT lens, though I am unsure what it’ll get me more than using distributed cognition (Hutchins) or just simply describing the learning arrangement between various humans and nonhumans to get the job done.

I guess my problem with ANT is that it seems boundless in terms of macro vs. micro analysis. As has been mentioned, an actor network can be made up of actor networks. Where does one start?

So, for example, I have a 40 person raid group that learns to kill a boss over several weeks. It seems like each person should be considered an actor that had to be translated into the network. We’ve also collectively used certain addons and tools within the game to help us manage cognitive load and to make transparent some of the underworkings of the game. Does each of these addons get counted? Does each iteration of an addon get counted (40 people running the same addon in slightly different ways, positioned on the screen differently, paying attention to different parts of the addon, etc.)? Do specific functions of the addon get separated as individual actors? Do different elements of the UI get separated? To back up, do specific people get broken down to mind-body-fingers?

Latour (writing as Johnson) briefly mentions that a door closer, an actor that’s been delegated the task of making a hole back into a wall, can be further broken down into the mechanisms in the whole object (egs. a spring, a metal cylinder). Is it completely arbitrary where a researcher draws the line?

In Reassembling the Social, Latour emphasizes tracing associations, which is possibly an answer to my above questions. I could concentrate on describing practice in the raid activity as I see it (which is pretty much what I’ve been doing for a while now), but pay particular attention to describing the functions of specific things as they relate to other things. Do this as they come up. In turn, these associations lead to other things that come up. Is that no longer considered ANT but after-ANT?

Is it more useful to describe cognition and memory and material resources within an entity a la dcog than use ANT? (Though my prob with dcog is more that it seems like a snapshot-in-time where I am trying to document the change in practice. ANT seems like it inherently considers instability and change through the act of translation.) Is ANT reserved for bigger arguments about societal relationships? About translation being the leveraging or convincing of other actors to share tasks? Or maybe a dcog analysis is the way to use an ANT lens using my ethnographic mehod…

Lots of questions. Maybe better suited to a blog post, as I’m just throwing ideas out there without much experience with ANT and such… But I thought I’d throw them out since it seems to that me the fastest way to learn something is to make transparent what you don’t know. And my digital ears perked up when I saw Tamara’s first message in this thread. ANT and MMOGs!

thanks,
mark

NO ONE replied except Bonnie Nardi off list! :(

And even then, she gave me some good pointers to articles I should read without any editorial comments of her own. Gah, more reading! :p

Was it not clear enough? I don’t explain distributed cognition at all. I don’t explain ANT at all because I assume the people who were talking about it know more about it than I do. I don’t explain WoW raiding, either, but I thought they’d all know what I was talking about. Also, I didn’t want to make the email even longer than it was…

Ah well… I guess I’ll keep reading.

There are six SLCs:

1. Center of Excellence for Learning in Education, Science, and Technology (CELEST) – most brainy
   
2. Learning in Informal and Formal Environments (LIFE) – most “everything is about life, dude”
   
3. Pittsburg Science of Learning Center (PSLC) – most original name
4. Spatial Intelligence and Learning Center (SILC) – most visual
5. Temporal Dynamics of Learning Center (TDLC) – quickest, yet slowest
6. Visual Language and Visual Learning (VL2) – most spatial

This past weekend the UW branch of LIFE (which also has branches at Stanford and SRI) hosted the second annual grad student and post-docs inter-center conference. It was pretty cool meeting all these other learning sciences students and hearing about their research. We were able to share tools and resources, findings, methods, theories and ideas, and some good drink and company at local bars after each day’s events.

There were a number of us interested in games for learning, from the use of virtual environments for studying the effects of 1st person vs. 3rd person POV on learning (Robb) to testing social vs. non-social feedback for navigation tasks (Dylan Arena), from task oriented vs. social oriented cultural learning goals (Amy) to collaborative activity-based multiplayer mouse control (Neema).

The first day, Sarah Walter from Stanford arrived early so we could meet and brainstorm proposals for upcoming conferences. She does almost the same research as me except that:

• I am focusing on trying to map the way a raid group works to an ANT or distributed cognition model where she’s focusing more on specific collaboration practices.
• My data only includes what players were already doing (chat logs, video, web forum threads), while Sarah’s got some interview and survey data in addition to what I’ve done.
• I’m looking at a 40-person raid in World of Warcraft, while Sarah’s group is a 12-person raid in Lord of the Rings Online.

We quickly saw that it would be easy to start using the same coding scheme and collaborate on analyses so we could compare our settings and findings. We’re writing abstracts to submit to IR10 (Milwaukie, Oct) and DiGRA (London, Sept). Prob will also submit to GLS (Madison, June 10-12) but she’s going to be at CSCL in Greece (lucky!) at the same time as GLS.

On Friday, we had a full day of poster sessions and then workshops on inter and intra center collaboration. We need a match.com for researchers, one that pushes info to participants when something new of interest (maybe tag based) gets added rather than depending on us to go visit a site routinely. Does that exist?

Afterwards, dinner at Portage Bay Cafe was pretty good. Met Vanessa who researches media realism and its effects on arousal.

On Saturday, we had presentations and workshops on current research and tools. The workshop I went to was the video analysis one and ELAN (presented by Sarah Fish and Naomi Berlove of VL2) looks great!

On Sunday, the conference was technically over, but I spent the day working at a cafe with Sarah Lewis (also from Stanford), lunch with Turadg, Erin, Ruth, and Ido (all from CMU), and working at a different cafe with Turadg. Sarah and I talked a bit about our programs and profs and politics. Very informative. :)

Turadg showed me some cool stuff he’s been working on that might help me with my chart creation… using python and pickling and a make file and such rather than going through all the crazy manual steps I’ve been doing with a text editor, excel, sql, flash, and photoshop. He’s also working on a collective web tool for learning that I’ve agreed to help with (though honestly, I only have a fuzzy image of what it is) and runs the Open Education Research blog.

Anyway, for anyone interested, below is the poster I presented (based off of the paper I was working on last month) at the conference. Also, you can get the bigger PDF version (13 MB).

Visualizing Chat Data in a WoW Player Group

Since the players in this raid group have all been playing for about a year and have reached the highest level in the game, they could be seen as expert players. This was not because they were experts of the game mechanics, per se. Rather, these players had successfully accumulated and displayed social and cultural capital, which depended on a fluency of the game culture above and beyond fluency with the game artifact (Chen, forthcoming (from the TWC article that I revised last month)). This research focuses on the adaptive nature of the raid group’s expertise, where the individual experts had to learn to coordinate and communicate effectively with teammates such that the group itself became an acting, thinking entity.

This would have let readers of the visualization paper know better what I meant by expertise and why I called the paper as an exploration of expert chat development rather than chat of the move from novice to expert…

Let me back up.

A few months ago (August or Sept) I was asked by Constance Steinkuehler to submit a paper for a special issue on games for eLearning. I didn’t really have anything I could write about with any sort of warranted claims, but I figured that I should take advantage of an invited paper plus it’s generally a good move to say yes to Constance. ;)

So, I emailed her telling her that I wasn’t sure I could make any strong claims as most of my analysis work has yet to be done but that I’d give it a shot. I spent a bunch of time exploring the use of visuals, namely charts, to look at the chat data that I have. This took a while, and given the time that I had, I decided to write a paper on how I made the charts (sent draft around Thanksgiving), since after doing a quick search I couldn’t really find any papers of that type that dealt with qualitative data in education. (There’s stuff from other disciplines and there’s stuff on quantitative data.)

I got feedback from her (Dec), mostly to make the findings and discussion areas more substantial and to move the “how-to” section to an appendix. In other words, to change the paper to not a methods paper but rather one on expertise development, which was the topic under study.

Edited and sent (early Jan), feedback given with a 24 hour deadline (two days ago), edited and sent again (and actually I think it is much better now and includes some chat data along with the charts to strengthen claims), but this morning I woke up thinking… “there’s not enough framing in the paper. I could have described what I consider expertise with these players better.” Oh well….

Claims? Based on analysis of two nights in Molten Core, separated by a couple of months, the first of which we encounter Ragnaros for the first time and the second in which we defeated him:

• the use of charts is very helpful but should be seen as complementing and supplementing deeper analysis of the content of the chat
• the level of overall talk rose but the cases of on-task talk remained at about the same level or even lowered. the thought is that as you become expert in something you can spend less of your time focused on the task and more on general banter since the task activity has become routinized
• women chatted less than men did. unknown reasons
• the time it took on the successful night actually lengthened rather than shortened, but most of it can be explained with the raid leader taking the time to explain the fight. we had become experienced enough to be able to talk through the fight before actual engagement

I’m revising that paper on visualization of chat logs in WoW raiding to map onto expertise development today and tomorrow. One comment is that the section on how I made the charts should be put into an appendix and the paper should focus on the actual argument rather than be a methods paper. That’s spot on, but I needed Constance to confirm it for me. One problem, however, is that some of my lit review is relatively sparse, but I can beef them up a bit, drawing from the excellent expertise work being done by my fellows (ESTG) at the LIFE Center.  Another problem, though, is my analysis isn’t as robust as it could be given that I haven’t had time to chart out all the raiding nights nor go into detail on specific nights to get a clear sense of exactly what is going on. But I guess I’ll have to do as much as I can in the next three days and hope it’s good/interesting/valid enough.

I played a lot of WoW over the holiday break, getting to level 80, getting pretty much the best gear I can get without raiding, etc. Now that classes have started and I’ve hit 80, I’ll probably cut back my WoW time significantly. My subscription is set to end in Feb… I don’t know at this point if I’ll be resubbing (again…).

In related news, I also tried out Tabula Rasa for a week or so. It would be much better with a regular group of people in a LAN party. As it is, it’s visceral and adrenalin pumping, but ultimately it feels like a grind and got too repetitive. It’s a free game right now until the servers shut down in Feb, so if you want to check it out (Windows only), let me know and I can play with for a bit…

I’m taking 4 courses this quarter rather than just working on my dissertation. All of these haven’t been offered during my stay here at UW before.

• Phil Bell and Suzanne Reeve (one of his students and one of my cohorts since I’m also one of his students) are coteaching a Learning Across Settings class, drawing from a lot of what LIFE does… specifically the “I” in LIFE (Learning in Informal and Formal Environments), internally known as Strand 2 of LIFE.
• Reed Stevens is teaching a Technology in Contexts class that will focus on Activity Theory, Actor Network Theory, and Distributed Cognition. These three theories are pretty much the ones I need to focus on in my diss, so this class seemed like a no brainer, especially since I get most of my learning from classes, meetings, face-to-face interactions, and conferences, not being a good independent reader and all…  too many games.
• Terry Schenold, a grad student in English, and Tim Welsh (who I haven’t met yet–our first class starts in half an hour) are leading another seminar on gaming this quarter as part of the Critical Gaming Project here at UW. This one is called Pandora’s Wake and is essentially visions of dystopic futures and hope as portrayed by Children of Men (film), Fallout 2 (computer game), and The Road (novel). Awesome. I installed Fallout 2 on my mini netbook this morning. Turns out there’s a few mods out there that fix bugs the original developers never fixed, add new areas and quests, and make it playable at higher resolutions. Due to compatibility issues (as in you can’t install all of the mods that are out there at the same time), I opted to go with the consensus as found on the No Mutants Allowed forums (killap’s Restoration Project and the Resolution Patch).
• And finally, I’m going to attend a seminar onthe college of ed’s common book, Lipsitz’s The Possessive Investment of Whiteness. Issues of social justice have been gaining prominence in what I’m thinking about, though, it hasn’t really surfaced in my research. This class hopefully will help me think through some things, as past classes, discussions with others, and involvement with the grad student group Educators for Social Justice has.

Anyway, all this work and new activity with the new quarter just starting up seems like complete bullshit against the backdrop of world events and the civilian deaths in Gaza. On the Red Square today were two protest groups standing at odds with each other on either side of the square, one group holding up signs like “Israel has a right to defend itself” and the other with signs like “As a Jew, I’m against Israel’s attacks against Gaza.” And meanwhile, we’re just walking through like this is completely normal. WTF is wrong with us?