When we found our first informant through a message board, one member of our group asked him if he could answer our questions regarding “trolling.” There was a long pause until he finally responded:
“Interview me is your first step. PayPal me twenty-dollars is your second step. Getting an A is your final step. Trolling is an art it’s hard to explain... It’s like robbing Van Gogh when he is in the Denny’s bathroom. Trolling is a form of expression like dance or writing music. The fanfare is to make people feel like they’re meaningless.” After that, we knew the project would be a long and arduous affair.
As the internet grows increasingly relevant in our everyday lives, it is important to understand the phenomena and cultures it is cultivating. One particular interest that our group decided to focus on is a behavior called “trolling.” The description of this nature is debatable, but through participant observation, interviews and research, our group was able to understand that this cultural scene resulted from “renegades” of these internet communities who disrupt and harass other users for amusement and/or notoriety.
What Counts As Trolling?
While setting down the basic outline of our research, our group often had trouble differentiating between different terms. Words such as “trolling, griefing, flaming, and spamming” are thrown around and used interchangeably at times which made it difficult to conduct research. Also, it was necessary to establish a common language to be used for interviewing. Prior to this, every time someone was interviewed about trolling, we had to first verify that their idea of trolling matches our description. This was a slow process; we often had to stop so that the person being interviewed could explain why each term meant what it meant to them. It was necessary to define trolling for the purpose of our research in order to expedite the process of gathering data. Prior to the interviews we would explain that regardless of what each term meant to them individually, we would be using the term in this particular sense.
Even though we established a common definition for trolling, we still asked what trolling meant to each person. The results often held true to our assumptions: trolling is when someone makes a post that is outrageous, hoping to get a rise out of other users for their own enjoyment. The targets vary, and the troll’s methods vary, but generally it’s making other people angry so that the troll can laugh at them and feel superior.
Oftentimes, the targets are “anybody who falls for it easily.” According to the people doing the trolling, these people are usually newbies to the site, of lower intelligence, have a superiority complex, or are trolls themselves. As one informants said, “Most of the time I am trolling I am doing it because I am bored and need something to do to amuse myself. I pretty much target anyone who displays a weakness which I can capitalize or is in a situation I can manipulate.”
Despite of what the informant above stated, however, there was a surprising find: trolls often troll other trolls. This makes trolls quite different from your typical playground bullies: instead of a group of trolls banning together to pick on a single person, many trolls explicitly expressed their distaste for ganging up on a single target. The trolls we interviewed did not like working together with other trolls and most enjoyed trolling trolls the most. When we asked why they did this, we had expected them to respond that it was fun as long as they felt like they were getting the better of someone else. However, the responses were overwhelmingly about status. These trolls trolled other trolls because it gave them more satisfaction to compete and win over an opponent of a similar level and mindset.
However, when we asked one person to clarify how this could be done, there was no distinct answer. Apparently, not knowing is fine as long as you think you’ve the upper hand and its fun. Thinking more on this issue, trolling trolls could very easily become an endless cycle, since so many different kinds of responses could be categorized as trolling. Troll A could post as a pseudo-naïve character, asking an outrageous question and expecting serious responses from people who don’t “get it.” Troll B comes in and tries to Troll A by responding in a furious and completely serious manner. Troll A will then respond thinking that they’ve found a sucker, and the cycle could continue forever. How would you know that you’ve won? For this question there was no distinct answer.
The question of when trolling occurs was met with some rather unexpected answers; I had originally meant the question to ask for specific times, such as right after someone tries to sound smart in a forum. However, those being interviewed took the question to mean something else and we realized our folly of expecting trolls to need a reason to troll. Most trolls troll when they are bored and want to have fun by messing with other people. Only one person interviewed said that they kept a specific alternate avatar strictly for trolling. Some troll only when they come upon someone who seems like an easy target, and will easily take offense, or there was already an ongoing troll attack that they choose to join in. Others purposely go looking for victims when they have time to kill, or start new threads in forums with outrageous headings to bait people into responding. Sometimes these trolls spend hours of their day continuously feeding this thread.
Contrary to what trolls would like to believe, their victims are not really offended for long. While trolls feed off of angry, excited responses, oftentimes once the victim realizes that they’re being trolled most of their anger dissipates. Rather than staying offended, most victims are just embarrassed that they fell for what the troll said as truth. Once it is clear to these victims that their attacker was just trolling for fun, and that they didn’t really believe in the things they said, they leave the thread quietly. We feel that this is the best and only thing people can do, since trolls get their laughs from making you post angry rebuttals.
Yes, there are still victims out there who know that they were trolled, but they are still deeply offended. These people tend to be perceived as taking things too seriously, and that only leads to be trolled even more.
Since websites thrive by having more people on it and using it, I was concerned over whether or not trolling would cause major damage. In theory, once there were enough trolls to hinder most people from using the site, the website would slowly die as more and more people leave. Most of those interviewed denied this would happen, including several victims. Trolling is not meant to incur real damage, and those who troll really believe that they’re not doing anything wrong. Even though their actions lead to their victims feeling less about themselves, trolls truly believe that they are righting the world with what they do. Alternately, when I asked whether or not the activity of trolls seriously hindered the intended activity on the website, most of my interviewees agreed that it did and that it was a real danger to the website that trolls were getting in the way.
Where Does Trolling Occur?
With the development of the internet, it has evolved into separate clusters, each with a multitude of forums where relevant topics may be discussed. As these hubs for information have flourished, so have those who would subvert these communities. Trolling is a text based harassment that has found a home primarily with in forum communities; however, not all forums, also called boards, experience this plight.
For trolling to occur, some basic requirements exist: the community must have a focus on text based communication, the community must also have the freedom to express themselves fairly openly without too much restraint (i.e. not predetermined responses), and it is typically large enough to have a degree of unfamiliarity between all the members.
By our definition of “trolling” as opposed to commonly seen negative behavior (i.e. griefing or flaming, etc.), we have determined that the desire to psychologically or emotional attack someone exists in text based communities, as other forms of interaction would add other incentives for this negative behavior. Without the freedom to express a person’s ideas we would not see the behavior we call “trolling” as with predetermined responses and created to limit the amount of communication and are generally geared toward a productive and positive community where negative behaviors and impulses are shunned.
Lastly, for the behavior we “considerer” trolling to apply a degree of unfamiliarity must exists between those involved. When this personal barrier does not exist, we see many of the negative aspects taken in a different light; such as an inside joke. When these comments and behaviors are treated in a humorous light, we no longer see the “attack” that we have defined trolling to fall under, so we do not treat those comments as instances of “trolling”. This is reminiscent to a passage Jennifer Mnookin wrote in a publication concerning the emergence of law in one of cyberspace’s first virtual worlds, LambdaMOO: “Remember, LambdaMOO is supposed to be fun. It’s a game. Can’t we all lighten up a bit?” A lash-back to new rules enforced by the administrations of the game, due to text-based harassment.
So long as these requirements are satisfied, trolling tends to exists in these communities; however, the frequency of trolling and whether or not trolling flourishes in the community have plenty of other markers. As trolls differ from one to the next, it is impossible to properly grasp each trolls habit; however, there are a few common situations in which trolling tends gather towards.
Seeking to incite responses out of a community, trolls tend to find a niche in forums that have strong feelings or forums that tend to have a bias regarding something. This “something” is vague precisely because it appears that forums for almost every topic exist somewhere on the internet; this “something” can range anywhere from gaming opinions, political views, religious beliefs, etc. A common method trolls use to incite a response in these situations is to simply place a comment or post that is opposed or opposite the general bias of a forum; many users that are associated with the forum will quickly reply to demonstrate their views and this situation is perpetuated. Another common situation that allows trolling to occur is the “question forums”. These forums are generally utilized for help in certain questions and the trolling of these forums has split into two broad categories: deserved or underserved. Underserved trolling applies to trolls who purposefully give useless advice (i.e. “Do a barrel roll!”) or comments merely to insult and put down the help seeker; this type of behavior is generally not accepted nor appreciated by most communities.
Deserved trolling applies in a narrowly defined margin, where trolling is applauded and appreciated because the question may have special circumstances attached to it; such as, questions that are forbidden to be asked as stated in the rules of a forum community or else questions asked for so often that a special post or sub-forum may exists to handle inquiries of those type. Trolls that attack these questions are accepted because it is perceived that members of the community that fail to conform and follow the rules are detracting and hindering the community. These two situations have developed into natural attractants for trolling behavior; however, just as there are natural attractants for trolls, some communities present a natural deterrent for this behavior based upon its structure. One prime example is smaller internet communities. As previously stated, trolls are more prevalent the larger the community. This reason cannot be solely attributed to the odds of interacting with the same members, because of open registration the ease in which a troll can create and “alt,” which means alternative accounts, for negative or positive purposes does not account for the absence of trolling on smaller communities. In some ways, it alters the presentation of an individual. In Nick Yee and Jeremy Bailenson’s writing, they posed the question: “Virtual environments, such as online games and web-based chat rooms, increasingly allow us to alter our digital self-representations dramatically and easily. But as we change our self-representations, do our self-representations change our behavior in turn?” The anonymity and “appearance” that was created for the sake of trolling most likely influences the attitude and behaviors of the individual.
Another heavily prevalent factor that helps maintain the level of trolling that appears upon various internet communities is the interactivity of the administration with the regular members of the community. In online forums where the administrators are not shadowy overlords dictating rules and presenting penalties, but utilize the forum as it was its original purpose, the sharing of ideas, those forum appear to have the least trolling activity. The two most plausible reasons for this lowered trolling activity are either the moderators are constantly dealing with the trolls as they will come across them in their own forum use, or trolls in such communities do not exists because of the fostered good will between all members and staff that appears to be created. Trolling appears to permeate all internet communities in a shape or form; however, trolls are far from the random attacks, rather they have a governing method.
For our original research, we used two methods to intimately understand trolls and trolling behavior. Initially, we interviewed three individuals with whom one of our team members are friends. Through these connections, a level of trust had already been established, so we sent a preliminary set of open-ended questions for the individuals to respond to via instant messaging. Since the questions were grouped together, the respondents answered our questions very quickly like a questionnaire. Also, there was little opportunity to immediately follow-up on their responses. Unfortunately, none of them have responded to our second set of questions which were composed to gather more in-depth responses. Our team member speculates that the friends did not perceive our research project to be serious work.
Secondly, we analyzed the text of an online forum where trolling occurs to better understand the context of these incidents. We briefly looked at the layout of the messages and the thread hierarchy of the website along with the additional features the forum might have such as indicators regarding the popularity of a particular discussion. Then we inspected the semantic organization of the website and compared its intended forum discussion topics to the messages posted on the forums. We identify the troll message as one that is completely off the topic of discussion or the topic of the moment that evolved from the leading thread. From the posting of the troll message, we read the responses proceeding it. Some of these messages were addressed to the troll, while some later responses were reactions to them. However, there were responses that were ambiguous to whom they were addressing.
Also, we attempted to contact trolls and a few respondents through the forums via the private messaging feature if it was available, which is usual on forums that require an account name. Unfortunately, after introducing our research project to our prospective informants, they did not respond. We speculate that they ignore the private messaging notifications for a few main reasons. The first major reason is that the message received notification is not displayed prominently when the user logs on. There was probably no improvement on this feature because it is rarely used. Second, the users may have read our message, however they may have felt that our message was off topic to the discussion. Some may have wanted to forget about the trolling incident and allocate their efforts towards their interests at the forum. Third, we did not gain enough trust with the informant despite guarantees of anonymity. Since we recently created new accounts at the forums, we may not have established enough of a reputation to identify ourselves beyond our message. With a lack of reputation around the forums, our intention for interviewing trolls may have seemed suspect. Lastly, the user may only created the account for a particular set of purposes: to troll or respond to trolling. Also, some users set up an account to troll over a set period time and then abandon it. So the user may not have seen our message at all.
It is difficult to define distinctions among flaming, griefing, trolling. Due to the growing popularization of Internet access and increasing unique users joining online forums, the definitions among flaming, trolling, and griefing have very little precision that distinguishes them from each other. This occurs as a result from the recent emergence of these phenomena, the lack of a central authority to define such a recent activity, and the multiple understandings of hostile online behavior. Although Wikipedia seems to emerge as a center of authority regarding knowledge about recent phenomena, there is a pervasive lack of reflection on the methodology of their definitions. Since these terms are undergoing debate and still gaining a consensus on their precise meanings, we will attempt to outline some criteria that might distinguish these activities. Julian Dibbell, a writer for the Wired Magazine, wrote, “Griefing, as a term, dates to the late 1990s, when it was used to describe the willfully antisocial behaviors seen in early massively multiplayer games like Ultima Online and first-person shooters like Counter-Strike (fragging your own teammates, for instance, or repeatedly killing a player many levels below you).” In some ways, trolling is a variation of griefing, except that it primarily uses a text-based form. The first criterion to consider is the number of participants involved in the activity. For all of these, the act can be performed by one user, but there can be a group of users that perform this anti-social behavior through signaling with the hostile message itself or coordinating outside of the site of the attack. Moreover, the number of those affected by the act should be considered. Flaming is believed to target only one user on a personal level, while trolling and griefing targets multiple users usually on an impersonal level. However, there can be cases where trolls and griefers seek to disrupt a particular user’s experience throughout the site.
Next, the site of the attack should be considered. Flamers use many online channels of communication, such as chatrooms and e-mails, to attack the user. On the other hand, trolling and griefing seem be contained in a forum or graphical virtual world, respectively. The last criterion to contemplate could be the level of annoyance that the affected users experience. An attack from flamers appears to be inescapable as they send continuous amounts of hostile messages from as many channels as possible, which disrupts the user’s ability to use those affected channels. Due to the impersonality of most trolling and griefing incidents, users can leave the certain parts of the site and ignore them. However, frequent repetition of those anti-social acts will decrease the appearance of social cohesion within those virtual worlds, and thus detract from their social utility.
More Trouble Shooting
A difficult issue that our research pertained to the subject matter; trolling is a behavior that is based off of wiles, mischief, transient users and deceit—many of our interviewees did not believe that we were being honest, and felt that we were trolls. In other cases, there were a lot of interactions that seemed fruitless. As one informant, nicknamed “Snapper,” wrote, “You already asked me these. S-stop. I am n-not a troll. Who else are you going to ask these to? If they’re a troll none of them will give serious responses.” One group member, who served as a field researcher, responded, “Like you?” After a short while, the informant began to cooperate and said, “Then.... they are not.... masters of the art. The art of “trolling.” I only troll defensively or people I don’t know. “The Art of Troll” ~ Tun (sic) Szu.” At which the researcher responded, “Alright, that’s enough.” It was difficult trying to cope with hours and hours of trying to find informants, and once that opportunity comes, the informant does not yield the results desired. However, after trial and error, we realized uncooperativeness was not necessarily “useless information,” as it is still a part of our data.
In some interesting situations, there were some interviewees who claimed they were trolls, because they were excited to have their names published in a paper. As one informant asked, “Will I be on this as KK51 or anonymous?” However, the researcher responded, “Sorry, but I don’t think it is ethnical for me to put your screen name on this. I have to cite you as anonymous.” The interviewee was disappointed and responded, “But I want to be on this as KK5.” Unfortunately, after this exchange, he ignored further follow-up questions. Again, while frustrating, this behavior was still a useful incite. Our perception of what was “useful information” or not was influenced by the fact that some of our group members had a myopic gaze, which means that some data that may be significant to a research might be lost because the researchers are already well-integrated in the field they are studying in. However, thanks to the use of detailed field notes, we were able to go over the data in detail.
It was interesting how there were some individuals who vehemently denied that they were trolls. While acting as a participant observer on a forum, the researcher noticed one member that acted extremely brash and insulted everyone on the forum in a ridiculous and exaggerated manner that resembled many other trolls she had previously interviewed. She sent him a private message and explained that she was conducting a research on trolls, and outlined in detail how he matched the description of a troll. However, he adamantly denied that he exhibited any behaviors. Before the conversation evolved into an argument, the researcher ignored him. It is unclear whether or not he did that because he wanted to troll our research, or because he honestly believed that the way he acted as fine.
Though the research was arduous, it was fruitful; however, because trolling is a modern phenomenon that is changing each day, we cannot ascertain that our definition of trolling and explanations for why it exists is absolute and universal. Arguably, we can make the argument that most individuals that engage in this type of aggressive and anti-social behavior are heavily motivated by jovial intentions, despite of the fact that their actions may seem extremely negative and brutish.
2008 Mutilated Furries, Flying Phalluses: But the Blame on Griefers, Sociopaths of the Virtual World. Electronic document,
http://www.wired.com/gaming/virtualworlds/magazine/16-02/mf_goons?currentPage=all, accessed March 19, 2009.
1996 Virtual(ly) Law: Emergence of Law in LambdaMOO. Electronic document,
accessed March 19, 2009.
1998 Where the Counterculture Met the New Economy: The WELL and the Origins of Virtual Community. Electronic document,
accessed March 19, 2009.
Yee, Nick and Bailenson, Jeremey
The Proteus Effect: The Effect of Transformed Self-Representation on Behavior. Electronic document,
http://www3.interscience.wiley.com/cgi-bin/fulltext/118533557/PDFSTART, accessed March 19, 2009.
Figure A—”Trollz,” Encyclopedia Dramatica
1 This username is a pseudonym for the purpose of this research paper