Comm, Coord, and Cmrdrie in WoW


This is the first draft of a paper I wrote a year and a half ago. I attempted to create a version for the web. Not completely happy with the UI and interactivity (lack thereof) in this medium… Originally, I was thinking of writing two versions of all my papers, but I gotta say, it’s a hellava lotta work.

This version is divided into pages and sub-pages. Some of the footnotes are found on the sub-pages which is why they aren’t all sequentially listed at the bottom…

Anyway, I thought it might be helpful for new scholars to see the different iterations this paper has gone through.

If you feel so inclined, you can check out the differences between all these versions… 🙂 This is all in an effort to make academia more transparent, goddammit.

A group of people who have banded together in the Massively Multiplayer Online Role-Playing Game World of Warcraft learned how to defeat the end-game dungeon Molten Core through collaborative improvement on communication and coordination over a period of months in an iterative fashion. I believe most groups who delve into Molten Core learn to complete it in a similar way. This particular group, however, was focused on sustaining and building player relationships and learning together rather than obtaining magical items or completing the dungeon as fast as possible. The group’s ability to reflect and be consistent about its desire to “hang out and have fun” allowed it to recover from a particularly poor performing night which threatened to disband the raid. Furthermore, camaraderie and joviality were clearly evident in chat among group members. In fact, on the poor performing night, the chat among members was relatively sparse and not cheerful. I argue that evidence of camaraderie is evidence of trust among group members and that the community of the group was able to successfully build trust among its members and rebuild trust when it was temporarily lost.

I aim to describe the communication and coordination practices of a group of players in the Massively Multiplayer Online Role-Playing Game (MMORPG) World of Warcraft by contrasting two nights of game playing. This group of players and I gathered twice a week to defeat the monsters in a high-end dungeon known as Molten Core (MC). We went through a process of trial-and-error with many, many failures before we finally succeeded in defeating all of the monsters in MC. Success depended on our group members’ ability to coordinate their efforts and maximize group efficiency by having each member take on a specialized role as determined by game mechanics and specific monster battles. To do this coordination, my group employed a variety of communication channels including specialized text chat channels for specific teams within the group. This particular group was also able to adapt and refine strategies and adjust to non-standard group compositions and non-standard character specifications. I argue that this group’s success was due to its members’ trust in each other and a shared goal of “having fun” rather than personal benefits from looting the treasure from the defeated monsters. I’ll further argue that this approach of emphasizing friendships might be a way to think about how people can be encouraged to cooperate and participate in other types of groups.

Theoretical Background: (Computer) Game theory [link]

World of Warcraft: the role-playing game [link]

Methods: Ethnography of World of Warcraft [link]

A typical night in Molten Core

Gathering and chatting

At about 5:15 PM server time (PST) on a Friday night in April 2006, my raid group started forming up, as it had been doing every Wednesday and Friday for the past 6 months. Our raid leader, Maxwell, was inviting the rest of us into the group, and I was invited early this night. Meanwhile, the rest of us were all over the game world—either as the characters who we bring into Molten Core or as alts working on other quests or pvping or whatever—or just logging into the game after getting home from work/school. Once we got invited, we knew we were supposed to make our way to the entrance of the instance, but getting everyone there so we could start took a while, as usual. Our official forming-up time is 5:30, and our official start time is 6:00, but we usually end up starting at around 6:15 because some people tend to show up late. That night we started fighting monsters at around 6:10. In other words, I was in this raid group for almost an hour before the group actually started fighting monsters in MC.

The task of forming a new raid group starts by finding enough people who want to go at a certain time. Once that is done (which in some cases can take months because friends want to be invited with each other and it is difficult to find a time which fits the schedules of 40 different people), the raid leader still has to deal with the task of getting everyone in the group together at the agreed upon time every week, twice a week. Some of us resent the fact that we sit around for upwards of an hour before actually fighting, but others don’t mind and use the time to greet each other and catch up with old friends.

We discuss new things about the game, new discoveries about the game, new strategies to try out, or otherwise engage in small-talk, and most of this talk is laid-back with a lot of joking around. For example, here’s a snippet of what the rogues were talking about that night while we were gathering:

18:00:46.484 : [Party] Rita: you guys have become familiar faces – I’m glad I’m with you all 🙂
18:01:04.734 : [Party] Thoguht: thanks! you too!
18:01:05.921 : [Party] Rebecca: hi Rita!
18:01:34.468 : [Party] Thoguht: We’ve been having some crazy rogues nights recently.
18:01:37.578 : [Party] Rebecca: what’s everyone’s best unbuffed FR?
18:01:43.234 : [Party] Rita: 137
18:01:52.468 : [Party] Thoguht: I feel lame.
18:02:03.734 : [Party] Roger: 92…
18:02:13.375 : [Party] Thoguht: I feel cool!
18:02:18.937 : [Party] Rita: I feel sexy!

Here one rogue, Rita, was just invited to the group that night. Then, as a way of greeting the other rogues who were in her party, at about 6:00 PM, she made an explicit comment about how much joy has come out of being part of our group. Rebecca and I responded and greeted back. I echoed that the last few sessions in the group have been really good to us rogues. What I meant was that both rogue loot has dropped and that we’ve had good success as a sub-group in the raid in terms of performing our roles well by dealing out good dps during fights and minimizing our deaths. Then, changing topics, Rebecca asked what each rogue’s fire resistance was. By talking to other players in other raid groups and reading strategies online, we knew that most people suggest rogues have at least 180 fire resistance during the fight with the last boss in MC, Ragnaros. When Rita said 137, I wrote that I felt lame because my fire resistance was low by comparison, but then Roger replied with a 92. I felt not so lame anymore (I had a fire resistance of 120). Playing off of my phrases, Rita said she felt sexy. This is a good example of the light atmosphere in our chat even when “on task” strategies and assessments are talked about.

Molten Giants
Pulling and coordinated fighting

After we all sufficiently gathered, we “buffed up” and started “pulling.” “Buffing” is the term used to describe the act of casting beneficial spells on other characters. “Pulling” is used to describe grabbing the initial attention of monsters who are found standing around at pre-set locations in the world. Once their attention is caught, they charge towards whoever did the pulling. The first encounter in MC is with two Molten Giants who guard a bridge into the rest of the dungeon. One could argue that 40 of us versus two giants is not really fair, and watching us fight them after we’d been fighting them every session for 6 months would support that argument, but, like most encounters in World of Warcraft, we had to learn how to approach the fight and what roles each different character class should play. For example, usually warriors are assigned “tank” duty where they draw and keep the attention (“aggro,” short for aggravation) of the monsters they are fighting so that healer classes can concentrate on healing the warriors rather than having to keep track of every raid member’s Health. The warrior class was designed to play the role of holding aggro effectively. They can activate abilities which are specifically for angering enemies and keeping aggro (Taunt and Intimidating Shout, for example)—abilities which other character classes lack. Meanwhile, healer classes—priests, druids, and shaman—can activate abilities or spells only they have which reduces some of the damage characters have taken. We usually have about five warriors in our raid group. Since most encounters in Molten Core involve just one or two monsters, we learned to designate two of our warriors to be Main Tanks (MTs), so that all of the warriors are not competing for aggro. The healers can then concentrate even more on these two warriors. Since we have multiple healers, too, however, we usually divide healing duty among them so that only a set of them are healing the MTs while the rest are either spot-healing the rest of the raid group when necessary or are assigned to heal specific parties in the raid. Monsters in WoW also have special abilities which they can activate against the players, and part of what we had to learn was the kinds of abilities to expect from each type of monster. For the Molten Giants, each Giant is tanked by one of our designated MTs. The non-healers, including the other rogues and me, focus-fire on one of the Giants until it is down. We then all concentrate our damage on the other Giant until it is also killed. Meanwhile, each Giant is concentrating on the MT that had been assigned to it, so those warriors are taking a lot of damage. The healers who were assigned to heal the MTs are healing them, while the other healers are healing the rest of the raid group who are taking damage from the Giants when they activate one of their abilities, a stomp which damages all surrounding enemies.

To aid us in this coordination, each role in the raid has a specialized chat channel. For example, the healers have a channel in which they manage the assignment of healing and buff duties:

18:21:48.843 : [3. healsting] Paula: how about Pod 1,2… Paula 3,4,5… and Peter 6,7,8? For DS buff

Here, the priests and other healers used the “healsting” channel. Paula was suggesting that each priest be assigned certain parties in the raid (there are 8 parties in a raid group, remember) to cast the Divine Spirit (“DS”) buff on. Only the healing channel is for multiple classes (the healing classes—priests, druids, and shaman), while each other class has their own specialized channel. This assignment of roles is common among all channels. Here’s an example from the warlock channel:

18:11:20.421 : [4. soulburn] Lori: Remember, ss target will change at Domo, but until then, your rezzer is to be ssed at all times.

Lori was reminding the other warlocks that an ability that they had to create a Soulstone (“ss”) and apply it on other characters should be active at all times. A Soulstone allows whoever it is applied on to resurrect him or herself. This is important to keep active on characters who can resurrect others (“rezzers”). In this way, if the whole raid group dies (“wipes”), our rezzers can come back to life and revive everyone else in the raid.

A lot of chat that happens in some of these channels is “off-topic,” featuring friendly banter. The earlier example of the rogues talking in party chat could have easily happened in the specialized rogue channel. It just happened that all the rogues were also in the same party, so we used party chat to talk to each other instead of using our specialized channel. This banter does seem to vary by channel and is probably due to the fact that there’s variation among people—some are talkative while others aren’t. It could, however, also be that some classes came to use the channels for certain things and got used to seeing their channel in that light.

While chat is happening in these specialized channels, concurrent chat may be happening in the raid channel, the party channel, the guild channel, and any channel that a particular player is subscribed to. Managing all of the information coming from these various sources can be challenging, especially when one has to concentrate on the virtual world and its encounters at the same time. In fact, reading through some of my transcripts shows pretty clearly that I missed some utterances which were directed at me. Also, sometimes the chat in one channel references chat in another channel. In this way, chat can be interwoven and layered. Furthermore, on top of the text chat, there is voice chat which is also sometimes running parallel to and sometimes interwoven with the text chat. Those who are not using voice chat can often see non-sequiturs in text chat. On the flip side, some people respond to the threads in a specialized text channel through voice which is confusing to those not participating in the particular specialized channel.

To get back to the story, to start off our night in MC, we pulled a couple of Molten Giants (after sitting and talking and gathering together for an hour). Our fight with the Giants was routine and only lasted a little over a minute. The text chat was relatively sparse because we all were familiar with the encounter and knew what to do. Even so, it was steeped in meaning. Here’s the chat from it:

18:11:34.671 : [Raid] Willy: INCOMING Molten Giant!
18:11:34.687 : Willy yells: INCOMING Molten Giant!
18:11:36.187 : Larry thanks Mary.
18:11:40.640 : [Raid] Lester: Pat is Soul Stoned
18:11:45.203 : Marcie hugs Lev.
18:11:45.562 : [Raid] Roger: rebroadcast ct please?
18:11:49.343 : Willy yells: ATTACK!
18:11:49.453 : [Raid] Willy: ATTACK!
18:12:57.359 : [Raid] Sherrie: This whole only shaman group is amazing!

First, Willy, who is the second-in-command, alerted the raid that we were pulling the Molten Giants. When this happened, the Giants charged our group and our two Main Tanks grabbed their attention. The MTs then ran in opposite directions and positioned the Giants so that the Giants’ “area of effect” (“AoE”) damage (their stomp) was not overlapping. This way we could kill one Giant without taking damage from the other Giant. While this was happening, Larry thanked Mary for something. I’m guessing Mary, who is a mage, gave some water to Larry. Certain classes, like mages and priests, can cast spells, like Fireball and Renew. Each spell they cast uses up a certain amount of Mana or magic spell points. They have a certain amount of Mana (depending on their character level and equipment), so after casting enough spells, they can run out and are no longer able to cast any spells. If they haven’t cast a spell for a while, they start to regain Mana slowly. If they are not fighting something, they are allowed to consume water, dew, milk, or other liquids to regain their Mana at a quicker rate. These drinks can be purchased in the game world, usually towns or cities, from certain vendors. Mages, however, can conjure water and share it with other characters, thus saving them from having to buy water. (Don’t ask where this water comes from; it’s best not to think about that.)

Next we see that Pat has had a Soulstone applied to her by Lester, so we had a safe rezzer in case something went horribly wrong. Then Marcie hugged Lev. In addition to Soulstones, Warlocks like Lev can create Healthstones and pass them out to other characters. Consuming a Healthstone will heal some damage, giving players a way to regain Health in an emergency during a fight if, for example, the healers have run out of Mana or if they are occupied healing the MTs. It could be that Lev had just given Marcie one of these Healthstones. It could also just be a friendly hug between two people who hadn’t seen each other for a few days.

Roger then asked if “ct” could be rebroadcast. Many World of Warcraft players use third-party addons which enhance or change how in-game information is displayed. Most of us use one called CT_RaidAssist which, among other things, allows raid leaders to designate Main Tanks. Once designated, little windows showing who the MTs are and what the MTs have targeted appear on everyone who used CT Raid’s screen. The CT Raid addon works by using its own specialized, hidden chat channel. Anyone who uses CT Raid would automatically be subscribed to that channel so long as the raid leaders synch everyone up by broadcasting in raid chat a certain key phrase which CT Raid recognizes. Players who join the raid group late or who get disconnected and reconnect often have to be resynchronized by having the raid leaders rebroadcast.

Then about 15 seconds after pulling and separating the Giants and then letting the MTs build up aggro, Willy called the rest of the raid group to attack. It took us about a minute after that to kill the Giants, at which point Sherrie announced that she liked being in a shaman only party. Shaman can place (“drop”) totems on the ground which give some sort of benefit to party members standing near them, but each shaman can only drop two totems, so they often have to weigh the pros and cons of which totems to drop. By having 5 of the shaman in one party, they were able each drop two unique totems for a very effective combination of benefits.

Making encounters routine by finding balance

After this fight, we prepared for the next pull by making sure our casters had regained Mana and that people were healed. The next fight was with another kind of monster which had different abilities, but it was just as easy for us with little danger of failure or of having lots of people die. In fact, our Molten Core experience has become a series of routine fights where we got ready, pulled, and killed in a systematic way until we reached a boss. These monsters were made so routine as to be called “trash mobs.” They are “trash” in that they don’t pose a threat and the loot they drop is often worthless in terms of making our characters more powerful but can sometimes be sold for good in-game currency (gold). This loot is also known as “vendor trash.” The term “mob” stands for monster object, which is how developers of MMOGs refer to game-controlled monsters or enemies.

Molten Core map

To make these trash fights a routine activity took us several weeks. For me, a rogue, it took time finding the right balance between doing a lot of damage (known as “dps,” derived from “damage per second”) and not taking aggro away from the tanks. The problem was that if I did too much damage (my dps was too high AKA I was dpsing too much), the Giant or Lava Annihilator or whichever mob we were fighting would consider me its greatest threat and start attacking me back instead of paying attention to the warrior who was tanking it. As soon as this happened, in most cases, I died. And this happened to me often, early on. After 6 months, one or two of us still have a difficult time of finding that balance, and it has happened to just about everyone in the raid group at least a few times. Even healers drew aggro by healing the warriors. The monster suddenly considered a healer more of a threat than the warrior in front of it. If enough of us did this—attracted the attention of the mob we were fighting—during a single encounter, the monster would “bounce” from person to person, moving to and killing whoever was the next highest threat. When this happened, usually an event known as a “wipe” would occur where enough of us died that there was no hope of defeating the mob before it killed the whole raid group. Learning each encounter involved many wipes, and when it happened, it took time for our healers to resurrect themselves and then everyone else. If we didn’t have any safe rezzers, we all had to release our “ghosts” in the game at the nearest graveyard and then run back to the entrance of the dungeon to reclaim our bodies and reappear in the world. While it can be frustrating to wipe over and over again, many of us in the raid, including the raid leader, took this opportunity (the time it took to either rez everyone or run back to the entrance from the nearest graveyard) to reflect about what happened and suggested things to change about our approach or suggested completely new strategies to try.

Welcoming failure in Golemagg and other boss fights

Since this night was several months into our raid instead of when we first started, we didn’t wipe on trash mobs. We also weren’t wiping on the early bosses. Our goal this night was to make an attempt on the last boss in the instance, Ragnaros. The way the dungeon is set up, our raid group had to kill all the other bosses before Ragnaros’ lieutenant, Majordomo Executus (“Domo”), would appear. Then after we defeated Domo’s guards, he would teleport away to Ragnaros’ chamber and summon his lord. This was a Friday night, so we had already been in the instance once this week and had already cleared out some of the dungeon including many of the early bosses, but we still had to defeat a unique Giant named Golemagg and his two Core Hound guards before reaching Majordomo. Boss monsters are special ones with more Health and more abilities. They often have minions or guards near them, and challenging a boss in these cases was a matter of tanking each guard along with the boss then figuring out which ones to kill first.

Lucifron early on - see all our dead bodies?
We reached Golemagg a little after 7:00 PM, about an hour after our first pull and about one hour and 45 minutes after we first started forming up for the evening. That is, we spent a good chunk of time just getting to a significant fight. Being a member of a raid like ours required patience and free time. Our strategy for Golemagg was to kill him before his Hounds because, once he was down, his Hounds would automatically die, too. To defeat Golemagg meant we had three warriors assigned to tank him and his two Hounds. While some healers were keeping the tanks alive, everyone else focused their attention on Golemagg. Golemagg has an ability which does periodic damage over a certain amount of time (“damage over time” or “dot”) and he can apply this effect on anyone within melee range multiple times. A rogue’s role is to run in, hit Golemagg a few times, run out of melee range when he or she has received enough dots, wait for the dots to wear off (because applying bandages could only be done when not receiving damage), bandage or otherwise heal (e.g. with a Healthstone) him or herself, then run back in to do more damage, backing off as needed. Again, learning the encounter was a balancing issue for rogues, maximizing dps without getting too many dots. If I stay within melee range to raise my dps a little, I might receive more dots than I could wait out. They would kill me before wearing out, preventing me from applying bandages. Learning the encounter for the raid meant we had to know the overall strategy of concentrating on Golemagg. We knew this because some of us had been in a fight with him before with different raid groups and some of us had read strategies online for the bosses in MC. Each class, like the rogues, has its own balancing act to perform since each class has its own set of abilities, all with different pros and cons. Golemagg has a plentiful amount of Health, and, this night, killing him took us almost 8 minutes (by contrast, the two normal Giants earlier took us a little over one minute). In long “endurance” fights such as this, it is common for healers and other casters to run out of Mana. If enough of our healers run out, the warriors are no longer getting healed. They would die, causing the rest of the raid to die soon thereafter since all the other classes cannot take more than one or two hits from Golemagg. The first few times we did this fight, like the first few times we did any of our boss fights, we wiped. This was not seen as a bad event but rather a necessary component of learning the strategy and finding the balance or “groove” needed to succeed. A raid member, commenting on a different boss fight, put it best:

Now I hope no one’s getting frustrated. This is how raids go. It’s normal: You fight and fight and fight until your gear is broken, repair and do it again. Once you finally get it down you can farm them for loots. It can take a while to master these encounters but we’re doing good work!

panther zerg - the fight being talked about in another dungeon where we died lots

Each time a character dies, his or her equipment suffers a durability loss. When enough deaths happen, the equipment breaks and can no longer be used. Repairing equipment requires a trip to a blacksmith or other vendor in town who can repair items for gold. This raid member was reinforcing the idea that dying over and over again, to the point of having equipment break, was normal and no cause to become frustrated. It took time and many attempts but eventually rewarded us with loot. Another raid member said:

Ultimately each of us can only control our own character; so the most important job we each have to do is make sure we are doing our part both effectively and efficiently… [S]moothly executing a kill on a boss that used to kick our tail is very gratifying, I think. ; )

For this person, the sense of accomplishment is very gratifying, and I think most members of the raid share his sentiment. It is not just loot we are after. We enjoy the challenge and success that comes with the hard work of failing multiple times. To succeed, each of us had to learn to play our role effectively.

Putting it all together for Majordomo

This night, we killed Golemagg relatively easily and moved onto Majordomo. Domo is an interesting fight which requires a lot of coordination. He has an entourage of four warrior-like guards and four priest-like guards. For this fight, unlike the Golemagg one, we want to kill the guards first. Once this was done, Domo would teleport away to Ragnaros’ chamber which was a necessary event for us to fight Ragnaros. Typically, we had five warriors in our raid group (this night we had six), so we couldn’t tank all of the guards and Domo at the same time. Luckily, the priest-like guards, unlike most other monsters in Molten Core, are susceptible to certain abilities of various classes known as “crowd control” (“cc”) abilities. These abilities effectively remove monsters from combat for a short duration. For example, rogues can stun certain monsters so they don’t fight back for a while. Mages, with a polymorph spell, are able to turn certain mobs into sheep which then wandered around, out of the fight until the spell wears off or they take damage, and this is the strategy we use.


In summary, two warriors were assigned to tank Majordomo (on typical nights when we only had five warriors, only one would be assigned to Domo). They had two hunters assigned to stand near them. If the Main Tanks lost Domo’s attention to another raid member, the hunters could quickly shoot a Distracting Shot at Domo to make him come back to where the tanks and the hunters were standing. Then the tanks could regain aggro on Domo. Meanwhile four mages were assigned the task of sheeping the four priest-like guards. Each mage had a hunter who would pull a guard to their location which was away from the rest of the fight. This prevented the sheep from unsheeping prematurely from collateral damage. Two of these initial pulling hunters were also the hunters assigned to the MTs meaning after pulling the guards over, they would run over to where the MTs were standing. Four other warriors were assigned to tank the warrior-like guards (on nights when we only had four warriors, a druid or shaman would fill in one of these tanking duties). One of these four warriors was the Main Assist. All damage dealers would assist this warrior in killing his target. He would then move to a priest-like guard, and we’d all kill it. We would kill all the rest of the priest-like guards then work on the other three warrior-like guards. The reason why we kill the priests after the initial warrior is that after four of these eight guards die, none of them are susceptible to cc anymore. The reason why we kill a warrior first is that we usually don’t have enough tanks to tank them all while killing a priest. Coming to this 1-4-3 strategy took some time, though. I remember trying a 2-4-2 strategy once when we were first learning this fight, but we wiped in a particularly nasty way on that attempt.

While these assignments were being made by the raid leader, the healer channel was occupied with healing logistics, the warlock channel with Soulstone logistics, and the rogues were talking about poisons which they can apply to their weapons to cause extra damage on certain monsters. Clarifying questions were happening on the warrior and hunter channels. Most of these specialized channels had a bit of chatter and commentary going on, and there was a good amount of joking around on the healer channel. In my transcript, I personally was also receiving messages from my guild, most of who were not in this raid group and from the officers in the guild talking to each other. Also, I was whispering or sending messages to various warlocks to get a Healthstone from them. Since I was subscribed to all of these channels, I could see transparently what was going on (provided I could actually follow it all), but to most of the raid, a lot of this talk was happening without their knowledge. I think this fact speaks volumes of each class’ trust in the other classes to play their part effectively. The warriors trusted that the healers were making the healing assignments, and everyone trusted that the warlocks were Soulstoning safe rezzers. During our fight with Majordomo, he activated abilities once in a while which bounced either physical damage or magical damage directed to his guards back at the attacker. This meant either rogues and hunters or mages, warlocks, and other spell casters had to periodically stop attacking. (Rogues generally took this time to bandage.) When one of these abilities was activated, CT RaidAssist would announce it on-screen and our raid leader would announce it in voice chat. Each raid member was paying attention to personal Health and Mana, the status of any potions, Healthstones, bandages, or other consumable items, the status of any class abilities which could be activated periodically, the text and voice chat that was happening, and details about the fight on-screen. Often, when I died, it was because I was not paying close enough attention to my Health, and the death could have been prevented by activating an ability or consuming a potion or Healthstone. This night was a “good” night. We killed Majordomo Executus easily.

We then moved, however, onto three failed attempts at Ragnaros. He proved frustrating because his encounter became “buggy” where he was activating abilities at odd times. We eventually gave up, and by the time we were done for the evening, it was almost 10:00 PM. Our gaming session was almost five hours, and, other than Ragnaros, was relatively successful. Golemagg and Majordomo Executus were effectively on farm status, and we got to Ragnaros without any wipes or even any danger of wiping. (Two weeks later we defeated Ragnaros for the first time.)

An atypical night in Molten Core

Combination of unfortunate circumstances

By contrast to our good night that Friday, the following week, we had an atypical night in Molten Core. It was atypical in that a series of events unfolded which I believe caused us many wipes and gave us generally poor morale which culminated in a “melt down,” where enough raid members fervently opposed each other on an issue as to cause strife and people quitting for the night. I think it started with having enough people in the raid be stressed about other things happening in their lives. We also decided that night to try using two different warriors as our Main Tanks for the first time. It was clear that the warriors who were not used to tanking weren’t sure where to position their monsters and that the warriors who were normally our MTs didn’t know which abilities they should be using and which weapons they should be using. To add to this, we had an abnormal group composition that night, with more shaman and hunters and fewer warlocks and rogues than we were used to. Though our raid did not strictly proscribe the exact composition of our group, it was still a combination of characters classes that we were not familiar with. This uncertainty manifested itself in our chat.

At various times in certain specialized channels, raid members were bickering with each other about where they were standing during some fights or doubting the role other classes were playing during fights. In other words, there was a distinct lack of trust this night. We ended up wiping three times on trash mobs. After our third wipe, no one said anything in text chat for eight minutes. That is, no chat was happening in the raid channel, none in the party channel, none in the say channel, and none happening in the various specialized channels for eight whole minutes. The longest idle time from our typical good night was two minutes. I think those who weren’t already feeling less than 100% became stressed out from our three wipes and the bickering that they were seeing in their specialized channels. At one point, the raid leader asked the raid if we should continue. We decided to continue which in hindsight was a mistake, because a few minutes later we had an argument break out over loot rules. This argument proved a shock to many of our raid members. Some heated exchanges took place over voice chat, followed by some heated text chat exchanges. It ended with some people, including our raid leader, retiring for the night.

I think for many of the raid members, it came as a shock since they did not see the entirety of the chat that was happening in the various channels. It also came as a shock to me because I was not paying as much attention as I should have to the chat while it was happening. I was dealing with some particularly stressful situations myself. I think this was similar to Barron’s observation that successful groups often were more able to manage their attention to their discourse on problem-solving strategies (Barron, 2003, p.332). The following day, many of us discussed what happened on the raid’s web discussion board.

The reason I stay with this group of players is not because I have a good chance to receive rogue loot (hah!) nor is it because I needed access to a group to do research. I honestly believe I have found the most caring, laid-back, and reflective raid group going to Molten Core. The events that happened that night were seen as a fluke. One raid member said, “I personal[ly] find what happened tonight to be just plane [sic] old rotten luck. We had a bad run tonight and people where [sic] getting tired and a situation accrued.” In light of this view, players were emphasizing the family nature of our raid group and how it is natural for people to sometimes disagree with each other. Another player said:

I love our raid. I know we are all going to get burned out at times and frustrated and upset and disagree with one another. It is part of being human. We are like brothers and sisters really. Stuff like this is going to happen. However I think we have all been playing long enough to know that we have a pretty great group of people going here and truly we care about and try to do what is best for one another.

We also talked about how we should treat each other in the future. One raid member said, “Stress, it happens. We have a wonderful group of people here and we should always keep in mind that every last one of these people has feelings.” What mattered most was that we learn from this experience. In other words, the raid group was treating this as cause for reflection by trying to identify the problem (or at least symptoms of it) and solve it. Personally, I suggested that we needed to consciously make the effort to lighten the mood:

Here’s my theory… I noticed that not many people were actually joking around with each other like we normally do… I think a lot of us were sick or tired or having a crappy day and when we got together we had enough people who weren’t feeling 100% that it showed itself in chat… and it showed itself in our performance and it showed itself in our stress levels.

it might seem artificial but if I notice that happening again in the future… I’m going to start making jokes.

Another raid member echoed my sentiments:

I also noticed the lack of joking around in raid chat, and vent was totally silent for the time i was on it. I agree hun…I will be right there with you making a nerd of myself to try and lighten the mood =)

To sum up, I think our lack of camaraderie was an indication that many people in the raid were feeling stressed more than usual and that some of them did not trust themselves or others to play their roles in the raid effectively. I also believe, however, that even though camaraderie is just a symptom of an effective raid rather than the cause of effectiveness, one way to “fix” a poor performing raid due to wavering trust in itself is for members to attempt to lighten the mood and be supportive of each other when trying new things. The raid group I am a member of also believed this to be true.

Ragnaros down after months of hard work
Issues and Conclusion

The raid’s general mood on any given night can be affected by just a handful of players’ attitudes as expressed through their chat. This is what happened on our “bad” night. I think the raid collectively momentarily lost track of its goals, but it was able to reaffirm them on the web forums the day after. These goals are friendship and “having fun” over the more traditional purpose of receiving loot. The raid’s realignment with these goals after a bad night was done through reflection and the ability to see that it had strayed while making suggestions for finding the path again. In a sense, the raid was metacognitive (Bransford, Brown, & Cocking, 2000) able to assess itself and determine how to get where it needed to be in order to reach its stated goals.

The raid itself does not think or act, of course. The raid is made up of 40 different players on any given night, and it is those people who think and act. It is difficult to say whether everyone in the raid felt the same way, and, in fact, I think it is clear that they don’t always agree; otherwise there would have been no strife. Yet, I think the majority of members feel very strongly about the familial nature of our group. This was especially clear when we debated how loot should be divided on the web forums during the early months of our raid. Rather than use a traditional point-bid system known as “DKP” (Wikipedia, 2006) that most raid groups use, which sometimes encouraged competition for loot, we use a weighted roll system, which allows committed members an advantage in winning loot yet still highlight the randomness of a roll system to emphasize the lighthearted nature of the raid group and it’s laid-back attitude toward loot.

Looking at game mechanics and systems to guess how players will behave can lead one to suppose that changing the rules of a game can encourage cooperation within situations that resemble social dilemmas. Actual player behavior, however, is complex. The concept of social dilemmas cannot model all of the different social aspects that go into the choices players make in their situated experiences. If one were to look at these decision-making points not as a series of rational choices but rather as points where players act out of emotion and role-playing, it becomes clear that the reason people don’t make group benefiting choices is that they don’t trust others to do the same. Trust happens to also be a key factor in the performance of a raid group working to defeat encounters found in World of Warcraft’s Molten Core dungeon. I think the trust needed for raid performance is the same trust as needed in loot division or other situations which resembles social dilemmas. The raid group I am in was able to foster this trust in its members by ensuring they were “in it” for the group and “having fun” rather than for individual, self-serving loot collection. It ensures this trust first by only recruiting players who other members had already established a friendly relationship. Second, the raid group explicitly states its goals in in-game chat and in the web forums and then reflects on its behavior in relation to these goals. Finally, the raid loot rules were collaboratively decided upon through its web forums, a process similar to one of the key components Kollock and Smith (1996) claim is needed for creating a sustainable online community.

The approach this group took may suggest a way in which teams in other settings (like work or school) can also take when working on a new task. Rather than focusing on the goal of doing the task right and reaping the rewards, teams can concentrate on building friendships and learning how to complete the task together. I believe, by fostering trust among group members, this process actually leads to a more coordinated group which is better prepared to handle future tasks and changing situations. Additionally, I think the way in which this group’s members valued friendship points to a possible solution to common social problems of trust and cooperation (Felkins, 1999).

References [link]

“High-end” here means that the game content was intended for players whose characters have reached the maximum level in the game.

Footnote [2] can be found on the page that it is referenced.

When characters took or dealt damage, the damage was of a certain type, one of which was fire damage. Along with building up resistances to the other types of damage, characters could acquire items which protect them from fire damage.

Grabbing aggro from the MTs and dying in such routine pulls is now met with laughter and people who do it are only jokingly chastised. Some even feel a bit of pride when it happens because it means they are out-dpsing others in the raid.

“Farming” is the term used for when certain monsters are killed over and over again for the loot they drop.

I’m using more of a social metacognition rather than an individual one. The raid members were able to define goals and assess themselves as a whole in relation to those goals.

In DKP, players receive points for each session in a raid group. They accumulate these points by going over and over again. When loot drops, players who want it bid on the loot using the points they have accumulated. Talking to someone in a raid group who uses this system, I learned that sometimes the members of his raid group would artificially raise the prices of loot by bidding on items they did not want, thereby forcing those who wanted the items to spend more points to get the item. By doing so, when an item came up which they actually did want, their competition could no longer place a high bid.

10 thoughts on “Comm, Coord, and Cmrdrie in WoW”

  1. This is from Zarrah, a fellow raider, posted originally at

    A good chunk of the paper is spent laying out the principles of a WoW raid. Presumably you're going to get multiple papers out of this. You might want to create a standard appendix you can just attach rather than laying out how a raid works for every paper. This would also allow you to explain your naming terminology and other conventions, and provide a glossary.

    This would also let you expound more on the interaction of our raid (and perhaps generalize about raiding in general), which is the interesting part. It would also let you discuss the motivation of our raid, as compared to other raids (some of which are loot orinted, held together with very strict rules, and highly regimented in order to keep people focused and possibly to minimize loot drama, and others of which are focused on the competitive nature of server- and world-firsts). Boss strategies can enter into this. For example, I know of at least one Silver Hand MC raid which puts everyone on the rune for Geddon. This means if a single person fails to run out when they are the bomb, they can wipe the entire raid, leaving almost no room for error. I personally believe they do this as a wakefulness check to make sure people are paying attention.

    Firefox gives a few errors in the formatting of the end notes, and I couldn't find end note #2.

    You may wish to discuss the level of interaction some people take. Some listen to vent but only communicate by speech. Some are primarily speech. Some start on typing, and listen, but eventually bother to become more involved by getting buying or making their microphones work, more fully integrating them into the raid, etc.

  2. Posted by Barberik, a guildie who sometimes went on the raids with me:

    Way better feedback than I had in a PM Zarrah, nicely done. Thog, I think a lot of your audience is giong to be MMO-familiar, at least.. the appendix/glossary idea is great.

    Your explanations are good, but they way they are interspersed, made it hard for me to stay on track and read through the whole thing, breaks up the flow of the prose a lot. And I’m interested as one of the subjects and someone who cared about your take on events described!

    Compare and contrast what you see different with the RL raid as suggested by Zarrah would be interesting if delved into more as well.

    So really, just posting to say “I second Zarrah.” Might be a different approach in a different version of the paper – e.g., for scholarly submission vs. more general web audience. Though if submission is to a ludological journal (are there any yet? I hope?) then for specialized audience probably also can drop a lot of the background explanation.

    Very cool to start seeing you write stuff up!

  3. More comments by Zarrah:

     Nothing like a couple of hours of yardwork to concentrate the mind. Warning: Wall of Text Inbound.

    The main point of thog's paper is that camraderie and trust can overcome poor performance when it was temporarily lost. That's a good point, but I think it needs a bit more punch. Specifically, what makes the RL raid different from other raids, and why is this important (vis a vis WoW and other groups).

    Yes, RL is different than most raids. I would like to see more about why. This may make for a longer paper than thog had in mind, but let me just outline some of my thoughts on the subject. Some of these are self evident.

    WoW, like most other MMORPGs, has a level cap, which means the only 'end game' progression is to improve itemization.

    WoW, unlike some other MMORPGs, has very limited ways of improving 'end game' itemization: you raid, or you do PvP. The best items in the game cannot be purchased, they must be earned in game. (Caveat: a sufficiently wealthy person could potentially 'buy' a raid or pay a service to level their character to High Warlord, but for the vast majority of us, this isn't going to work.)

    Therefore if a player wishes to materially improve their character (a natural state for most players, tho by no means all), they must either self organize into raids, or self organize into smaller PvP groups (as it's very difficult to make HWL/GM solo). This is the main reason most people raid (beyond the initial curiosity of seeing new content). After you've killed Rags a couple of times, you've 'done' MC. You keep doing it either for more loot (Full Tier 1, T2 pants, other class-useful drops), because it's a stepping stone to more content (BWL), because you enjoy doing MC, or because you enjoy being with the people who you're doing MC with.

    Again, the general case (IMHO) is that people mostly do it for the loot, or because it's a stepping stone for more content. The upshot is that most raids are organized around loot and content, and wind up run in a very regimented way: DKP, set class roles, guild-only raids (which allows a raid to 'control' a given member more tightly than a mixed-guild raid, as leaving the raid or failure to perform can mean being booted from the guild as well as from their position in the raid, losing access to the raid bank and other services), required specifications, etc. I would really like thog to do some research here and outline a couple of different types. I also think that Leftovers would be a fascinating study here because they are so very different (and successful) than any other raiding structure I'm familiar with. Contrast them with the huge Death and Taxes thread from the Guild Forum, where you have a small (~50-55 raider), end-game-focused guild which is a combination of very regimented, but also very tight knit.

    Unlike Leftovers (which has a structure where anyone can Raid so long as they show up, do their job, don't cause drama, etc., and which seems to be loot and content oriented), and unlike a 'traditional' guild structure, RL is very different.

    The raid is cross-guild, which reduces the 'sticks' the raid leaders have to keep raiders focused on the job at hand. The raid is open to new members, not requiring particular specs, gear minimums, auditions, interface mods, vent capability, or much beyond a good attitude. Entree into the raid is informal, and the raid has fairly high week-to-week turnover, but it is effectively by invitation, particularly after the first time. In order to fit into the raid, you need to be both minimally competent at your role, but also be able to fit into the RL raid environment socially. The loot system, the single biggest general point of contention in most raids, is very informal. Thog brings this up but I don't think he spends enough time discussing why this is important.

    Let's consider that the RL raid can do six bosses in a 4 hour MC run. This is a minimum of 8 BoP epics (probably closer to 10), plus associated BoEs, greens, mats, etc. Let's call it a total of 12 epics, plus 500g worth of materials (cores, essences, etc.) and another 400g in flat cash. The greens, mats, etc. are effectively the property of the raid (and go for repairs, potions, enchants, or improved raiding gear). The 400g means each raider gets about 10 gold per run (seems high but I'm basing it on getting about 1.4g per boss plus trash cash), which doesn't quite cover the cost of raiding.

    This means that each epic takes 3.33 people 4 hours of work. At an assigned cost of $10 per person per hour, an epic is worth $133 in labor alone. This doesn't include raid leader time, guild officer time, farming time outside the raid for materials, etc. This is with a mature raid that normally one shots bosses and is looking at being able to condense all of MC into a 5-7 hour run. The labor cost per epic in a new raid is much, much higher. Thus, loot drama, and the reason for DKP and any number of other loot distribution systems that concentrate very hard on being 'fair', or oriented toward 'raid efficiency'; priests don't get +DAM caster gear, druids don't get melee gear, other gear restricted to the main tanks until they're equipped, and so on. (Consider that the next time we melt something for a nexus crystal, too.)

    Despite all of these things, the RL raid is successful (as in having longevity and a certain amount of success) where a lot of loosely built raids have fallen apart, and they do tend to fall apart. We hand these epics, which have a significant real world dollar figure attached, around with pretty minimal drama despite a fairly informal loot system. Why? That's the question I think thog is addressing, and I'd like to see it addressed in more detail. His example, of a raid gone wrong, is relevant because of how the RL raid handled a challenge faced by any raid. Nights like that can break a raid if people get disgusted and walk away. In a 'loot and content' raid, it pretty much falls to the raid officers to enforce discipline and keep people focused. In the case of the RL raid, you wound up with what bordered on a bottom-up response to a bad situation (coupled with a purposely built-in leadership redundancy, but that was more in addressing the mechanics of the boss fights, and not in enforcing raid discipline).

    There is a side issue regarding just how complex these raids are that he's discussing, and the fact that a good raid can get to be pretty much self-organizing. I think there's a lot of interesting study to be done here as well.

    There is another side issue regarding guilds and their purpose within WoW. It's pretty clear that the history of guilds in MMORPGs is built around high end raiding. In WoW, particularly on RP servers, they've become something different, and RL is yet again a good example of this, despite being (partly) a raiding guild. The fate of Afterlight would be a good study here.

    Okay, enough blathering. 

  4. Love the comments, guys. I'll be editing the paper over the summer based on as many comments as I can get from academia AND gamers.

    Human subjects note:  posting this paper to my blog… making it transparent and then inviting comments thus making it so that people who comment are not anonymous might be iffy in terms of my school's policies on research with human subjects… on the other hand, screw academia! this is a frikken blog! but anyway, post if you want your name out there, otherwise email me if you want to remain anonymous (though, technically, email isn't guaranteed confidential…pfft)

  5. Hi, I assume you mean how do you start a guild (GM = guild master)? If so, there are NPCs in all the major cities that you can talk to to buy a guild charter. You then need to get 9 other people to sign the charter.

    If you actually mean, how do you become a game master which is the customer support people that Blizzard hires to handle issues players have in-game.. well, I imagine you’d have to apply for the job.

    I suppose a third possibility is that you’re asking how to page a GM in-game if you have a problem. There is a button down near the Options button for you to do that…


  6. Pingback: dissertations
  7. Pingback: soulburn

Leave a Reply

sporadic ramblings of a gamer in academia