In July, Simon Ludgate wrote a feature about the three F-Words of MMORPGs on Gamasutra. One of them was Fairness (link). He concludes that if the MMORPG progression is based on time played, then no one really thinks about fairness. However, if the progression is skill-based, then people do get upset when the game allows skipping ahead by buying progression. He also states that if the progression is based on accomplishing challenges, then those challenges should be targeted to the largest possible player pool. Making challenge accomplishments tradeable, he ends, allows players to complete the challenges they want, and trade the accomplishments of the challenges they don’t care about, which is a good thing.

Without implying that Simon’s argument is flawed or incorrect in any way, I’d like to add to this discussion. The kind of fairness Simon talks about is one that keeps the player happy in terms of playing against the environment and against other players, but it disregards one problem: socializing. MMORPGs differentiate themselves from all other game genres because they provide ways for players to play together. MMORPGs allow people to group together specifically to go up against challenges that cannot be tackled by a solo player. And as in any other collaborative activity, some people will decide to see this activity as a vehicle for their own growth and enjoyment, even against the growth and enjoyment of the other members of the group.

Take World of Warcraft, for example. Players may take part in a group until they come across the loot they’ve been after, and then just leave. But this is only one reason. Players may decide to leave the group for various reasons. The end result however, is that the rest of the group is left helpless, and it could be a considerable amount of time until another player is assigned to the group (especially if the tank is the one deserting). This behavior exemplifies a problem that game designers face when building collaborative structures in a MMORPG.

I wrote an academic paper about this (Christou, 2011), noting that the implementation of sanctioning mechanisms as game rules may be a good solution to this type of problem. I created an online survey, and recorded 31 PUG instance run conversations in WoW. I then analyzed the data using chat transcript analysis on the records of the conversations. The survey, left active for approximately 9 months during 2010, consisted of 13 questions. 155 WoW players responded to the survey.

The data from the chat logs revealed three reasons for leaving a group: a real-life problem, instrumental behavior, and problematic group. A real life problem occurs when a player cannot continue due to a situation outside of the game. Instrumental behavior is displayed when a group member decides to leave the group because their personal aspirations from the instance run have been met. Finally, a problematic group occurs when the players as a group do not have enough power to defeat the instance content, resulting in continuous wipes.

Once players have been grouped together and enter the instance, they are expected to stick around and finish the whole instance, to clear it. This can be thought of as a social norm in MMORPGs. In fact, from my experience, players usually debate whether to compete against all the bosses in the instance, not only against the compulsory ones. Of course it is justified to deviate from this norm, and quit an instance run. This is where the survey results came in. To gauge the players’ feelings about quitting runs, I asked the respondents to choose the acceptable reasons to quit a run, from a list which was by no means exhaustive. The questionnaire didn’t include answers that discussed real life disruptions, as it is understood that real life takes precedence over a game.

The results showed that there are two main reasons on which a large number of players reach consensus.  The first was “bad player behavior” with 82% of the respondents agreeing on this. The second most popular reason was “more than 5 wipes on a boss”, with 60%. The least agreed upon reason to leave a group was “it is OK to leave the instance as soon as they have acquired enough points to purchase gear, or because one of the bosses has dropped the loot that you wanted”. Only 6% of the respondents answered accepted this as a good reason to leave the group. This is one type of instrumental behavior I mentioned earlier. In such cases, the group may need to wait for some time for a replacement, who is robbed of their chance to earn the entire number of points given during a run, or roll on the loot that was taken by the deserter.

WoW does use a sanctioning mechanism to alleviate this problem. The player leaving the group is banned from attending another pick-up group (PUG) instance for 15 minutes. This ban, called the “Deserter Debuff”, runs its course in 15 minutes of real time whether the player is logged on or not. 70% of the respondents to the survey answered that this was a good way of punishing the deserter. The way this action is sanctioned in WoW now is to apply the deserter debuff to everyone, regardless of the reason they leave. However, there are legitimate reasons for leaving a group, with a large consensus by the community on the matter. Also, 75.34% of the respondents answered ‘No’ to whether they would accept a player who quits easily in their group if they knew about the player’s behavior.

Respondents were asked to suggest other sanctions to players who display this type of behavior. Some of their suggestions were that the debuff’s time should be increased, its duration should be counted in play time and not real time, or any loot and points gained by the deserter during the particular instance run should be taken away.

So all of these statistics, taken together with findings from the literature on collaborative behavior provides some ideas about dealing with this phenomenon. The literature on collaborative behavior suggests that if defection is allowed, then the members of a group will cooperate for a little bit, and then that cooperation will degrade into instrumental behavior (Dawes & Thaler, 1988). But if a mechanism is provided to sanction defection, it induces a sense of security to all the members that the others will cooperate at high levels voluntarily (Fehr & Fischbacher, 2004).

Suggestion #1: Include a mechanism through which group participants can sanction the players that leave the group.

Now don’t forget that the survey participants also said that they would not accept a multiple offender in their group. This leads to a second suggestion.

Suggestion #2: Create a mechanism with which to show stats about completion rates (and even other things, such as dps, gear score, etc.) to the other players in the group before the instance run starts.

Literature on human cooperation also suggests that human cooperation is based on conditional cooperation. This means that “if the other group members cooperate then I will cooperate. If they don’t I probably won’t either.” This attitude makes defection easier for players (Fehr & Fischbacher, 2004). So if a group member deserts the group, a suitable replacement needs to be found soon, otherwise the rest of the group members will desert the group as well. WoW provides a tool that provides quick replacement of deserters, and I believe that it helped to “save” many groups that had deserters. On the other hand, before Blizzard had the tool in the game, finding a replacement for any role was an involved process.

Suggestion #3: Follow WoW’s example to have a tool that allows a group to fill-in the open positions in a group quickly.

As theory on conditional cooperation suggests that the absence of sanctions is likely to cause decreasing contributions over time from team members (Fehr & Fischbacher, 2004), if the sanctions imposed do not impact the sanctioned players, then the sanctions might as well not exist. In other words, with no sanctions, or weak sanctions, the behavior of norm-obeying players will change over time, and the norm may eventually become the exception.

Suggestion #4: Create sanctions in such a way that they impact the remaining players as well.

Deciding to include the option to vote a player off the group may sound like a good sanctioning mechanism to implement, especially in light of the previous answers to the survey that bad behavior is the most agreed upon reason to leave a group. However, this same sanctioning mechanism will work against inexperienced players, because if the group agrees that a player is “too weak” for an instance, and they decide not to help that player, then how will that player ever get the required experience and rewards to become “good”? This type of mechanism may create a vicious circle for inexperienced palyers that can only be broken through an existing social network in the game (Jakobson & Taylor, 2003).

Suggestion #5: Think about the repercussions on all players types before implementing your sanctioning mechanisms.

To recap, if we add formal rules that sanction “quitters” without giving the rest of the group some power over those sanctions, we may not deal with the problem as effectively as if we allow the group to decide on the punishment of the quitter. To make your game fair to all involved, add sanctioning mechanisms, but remember that people are quick to pass judgment on others if they have no repercussions (Anderson & Putterman, 2006; Ones & Putterman, 2007). On the other hand, if people who sanction others also incur some form of reduced payoff, then they punish others less.


Anderson, C. M., & Putterman, L. (2006). Do non-strategic sanctions obey the law of demand? The demand for punishment in the voluntary contribution mechanism. Games and Economic Behavior, 54(1), 1-24.

Christou, G. (2011). Sanctions, Punishment, and Game Design: Designing MMORPGs for Fair Treatment of Players. Paper presented at the GAMEON-ARABIA 2011, Amman, Jordan.

Dawes, R. M., & Thaler, R. H. (1988). Anomalies Cooperation. The Journal of Economic Perspectives, 2(3), 187-197.

Fehr, E., & Fischbacher, U. (2004). Social norms and human cooperation. Trends in Cognitive Sciences, 8(4), 185-190.

Jakobson, M., & Taylor, T. L. (2003). The Sopranos meets Everquest: Social Networking in MMOGs. Paper presented at the Proceedings of DAC2003, Melbourne, Australia.

Ones, U., & Putterman, L. (2007). The ecology of collective action: A public goods and sanctions experiment with controlled group formation Journal of Economic Behavior & Organization, 62(4), 495-521.