Chapter 3 -‐ Methodology
3.1 Hardaker’s Model of Trolls
國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
YouTube users incited PewDiePie’s fan community by posting insults towards PewDiePie and his fans such as “PewDiePie sucks,” or “PewDiePie’s fans are all 8-‐year-‐olds.”
There were many of these negative and hateful comments but I excluded comments which attracted little to no respondents. Instead, I selected comments which contained an ongoing conversation chain of at least 50 or more responses for two reasons: 1) I could weed out “haters” who posts negative messages but do not respond to anyone who has replied to their original comment and 2) to identify successful trolls who are committed to and have succeeded in engaging their victims for longer durations. In these comparatively lengthy conversations, the suspected troll must continue to interact with his so-‐called victims, extending the spiral of disruption and pursuit of lulz.
Other red flags, which indicated the presence of a troll, included hostile behaviors such as ‘flaming’ and ‘hating,’ as described in chapter 2.7; this may include insults, negative and/or vulgar language, and strings of capitalized text.
Furthermore, I was the only coder working on all analysis for this study so there was no need for inter-‐reliability tests and all results are consistent in that they were analyzed by one person. There are, however, issues that may arise from single-‐researcher analysis as explained further in the limitations section.
3.1 Hardaker’s Model of Trolls
I have chosen to analyze the interactions of YouTube trolls with other users
according to Hardaker’s (2010) characteristics of trolls in asynchronous computer-‐mediated-‐
communication (CMC). Hardaker’s study originally sought to redress the lack of a proper academic framework to describe the phenomenon of trolling. As a corpus which analyzed data from several year’s worth of unmoderated Usenet newsgroup posts, Hardaker’s study
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
looked primarily at how trolls were being described by community members, sometimes including the trolls themselves, and ultimately identified four recurrent themes regarding trolls/trolling: deception, aggression, disruption, and success.
Given Hardaker’s classification of Usenet’s user interaction system as asynchronous CMC, it is reasonable to infer that her findings are also applicable to other online forms of asynchronous CMC such as YouTube’s commenting system. YouTube users interact through computer-‐mediated commenting areas provided under each YouTube video and
asynchronously by posting messages to one another whenever they are online rather than in real-‐time. Therefore, responses may be posted within hours and days to months or even years in between.
Looking more in-‐depth at Hardaker’s characteristics of trolls, ‘deception’ refers to users who attempt to pass themselves off as sincere community members instead of overtly identifying themselves as a malicious troll. ‘Aggression’ is described by Hardaker as
“aggressive, malicious behaviour undertaken with the aim of annoying or goading others into retaliating,” while ‘disruption’ includes causing aggravation without necessarily attacking specific individuals. Finally, ‘success’ entails trolls being appraised by users
regarding a troll’s quality and how others respond to the troll. I posit that ‘success’ is tied to the notion of lulz, which remains the primary motivator for trolls and the entire basis for troll culture. I hypothesize that trolls themselves may also speak about success or defeat in their communication with others.
I intend to test Hardaker’s model of trollish characteristics, aiming to seek if and how these four characteristics manifest themselves in the interactions of YouTube trolls and other users. Acknowledging that Hardaker’s model was created from ground level research, and may not be fully compatible with study on YouTube trolls as she analyzed UseNet trolls,
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
I will also attempt to account for any differences and identify any additional characteristics prevalent within YouTube trolls that were not mentioned in Hardaker’s study. For example, Sierra’s (2014) notion of “Kool-‐Aid Points” for trolls who attack others because they feel their attention and status is unwarranted, might be a motivator for trolls to target PewDiePie’s page. Thus, I propose the following research questions:
RQ1. Do YouTube trolls exhibit the characteristics of 'deception,' 'aggression,' 'disruption' and 'success'?
RQ2. Are there any other characteristics that YouTube trolls exhibit?
RQ3. If so, how do trolls exhibit these characteristics?
There are a few points of contention that require clarification regarding the use of Hardaker’s model. Hardaker’s formulation of trollish characteristics is based on analysis of the Usenet community. Although her model of “trolls in asynchronous CMC” may be a reasonable starting point for analyzing YouTube trolls, there are bound to be certain idiosyncratic differences between YouTube trolls and Usenet trolls. Thus it is reasonable to expect areas in which this study’s findings diverge and/or contribute to Hardaker’s model.
Additionally, Hardaker evaluated descriptions of trolls within the Usenet community in order to construct a model of trollish attributes, whereas I aim to utilize her model to directly analyze trolls and their comments. While it is true that I may not be able to completely isolate trollish comments from other users’ comments, this study places higher emphasis on directly examining characteristics of trolls as exhibited by the trolls themselves, more so than indirectly examining descriptions of trolls by other community members.
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
Additionally, there might be a few hidden themes regarding the nature of trolls, as mentioned in previous sections. Existing literature has theorized that many trolls are indeed white young males who many times seek to direct their trollish behaviors towards women, minority groups, and homosexuals (A. Chen, 2014; Herring et al., 2002; Milner, 2013; Phillips, 2011, 2013). Ironically, “maleness and whiteness” have also traditionally been used to
describe gamer identity (A. Shaw, 2010) which belongs to the very culture which PewDiePie espouses through his videos. I will attempt to confirm or disconfirm if such themes do rise up during my analysis of YouTube trolls:
RQ4. What are the social implications of YouTube trolls?
‧
國立 政 治 大 學