A research report funded by the European Commission from Despoina Chatzakou, Nicolas Kourtellis, Jeremy Blackburn et al from Aristotle University of Thessaloniki, University College of London and Telefonica Research has been released. The report can be viewed over on Arxiv, and it’s titled “Measuring #GamerGate: A Tale of Hate, Sexism, and Bullying”.
The report starts by claiming that the researchers decided to examine #GamerGate due to sexism, harassment, discrimination and bullying. They state…
“[…] online aggression and abusive behaviors have occurred in many different forms and on a variety of platforms. In extreme cases, these incidents have evolved into hate, discrimination, and bullying, and even materialized into real-world threats and attacks against individuals or groups. In this paper, we study the Gamergate controversy. Started in August 2014 in the online gaming world, it quickly spread across various social networking platforms, ultimately leading to many incidents of cyberbullying and cyberaggression. We focus on Twitter, presenting a measurement study of a dataset of 340k unique users and 1.6M tweets to study the properties of these users, the content they post, and how they differ from random Twitter users.”
They draw a conclusion and then attempt to use the data to reach that conclusion, as opposed to neutrally examining the data for what it represents.
For instance, the report claims that #GamerGate is a harassment campaign, aggressively targeting women, but then their actual data shows that #Gamergate users don’t use the kind of tone that the data correlated to being aggressive language, such as “shouting” in all caps.
As mentioned in the report…
“ GG and random users tend to use emoticons at about the same rate […]. However, GG users tend to use all uppercase less often. As mentioned, GG users are savvy Twitter users, and generally speaking, shouting tends to be ignored. Thus, one explanation for this behavior is that GG users avoid such a simple “tell” as posting in all uppercase, to ensure their message is not so easily dismissed.”
Here’s another thing that makes no sense. If the objective is to harass people offline and bully them, why would GG users care about their message being “easily dismissed”? A bully isn’t there to convince anyone of their viewpoint, they’re there to bully. The report turns in an oxymoron response in relation to #GamerGate’s intentions and the actual data related to user activity.
The report hilariously continues on this path, where more data goes against their narrative about #GamerGate being a mob of angry, virulent, unruly, malcontent young men. They report…
“[…] based on , we extract sentiment values for different emotions: anger, disgust, fear, joy, sadness, and surprise. We note that of these, the 2-sample KS test is unable to reject the null hypothesis except for joy, as shown in Figure 5d. This is particularly interesting because it contradicts the narrative that Gamergaters are posting virulent content out of anger. Instead, GG users are less joyful, and this is a subtle but important difference: they are not necessarily angry, but they are apparently not happy.”
“Contradicts the narrative”? So real world, measurable data seems to show that #GamerGate isn’t quite what the media says it is? Oh, say it ain’t so Plucky… say it ain’t so.
Also, is it no surprise #GamerGate users are unhappy when the main complaint is that media is corrupt and the media just keeps digging in their heels with more corruption? Why would anyone be happy about media nationally and internationally posting lies, hate-bait and misinformation? No one is happy about that, and according to a Pew Research report, 70% of Americans dislike their own mainstream media for spreading the exact same kind of negativity.
They did note that #GamerGate users tended to use more words from the “hatebase” database, which collects and measures “hateful” words on a scale from 1 to 100. Average users score between 0.06 and 2.16 on average for using “hateful” words, where-as #GamerGate users scored between 0.25 and 3.55.
The report continues to contradict its lede, where they painted #GamerGate as an online mob campaign spreading hate and harassment given that they found that #GamerGate users actually tended to be suspended less than average users. This actually fits in line perfectly with the WAM!, peer reviewed report, which indicated that very few Twitter users associated with #GamerGate were reported in relation to harassment… only 0.65% of the users from the #GamerGate block list were actually reported for harassment.
As for the Twitter measurement report, it states…
“ From Table 1, we observe that, in both cases, users tend to be suspended more often than deleting their accounts by choice. However, baseline users are more prone to be suspended (20%) or delete their accounts (13%) than GG users (9% and 5%, respectively). This seems to be in line with the behavior observed in Figure 2a, which shows that GG users have been in the platform for a longer period of time; somewhat surprising given their exhibited behavior. Indeed, a small portion of these users may be spammers who are difficult to detect and filter out. Nevertheless, Twitter has made significant efforts to address spam and we suspect there is a higher presence of such accounts in the baseline dataset, since the GG dataset is very much focused around a somewhat niche topic.”
The biggest issue with this particular report is that the entire first half is filled with an editorialized bias against those who used the GamerGate hashtag. This resulted in a rather embarrassing display of the correlated data contradicting what was purported in the lede.
They should have simply stated that they were examining the data and let the numbers speak for themselves, similar to the WAM!, report.
The piece concludes with a stark lack of self-awareness about the data not matching the narrative. They state…
“This paper presented a first-of-its-kind effort to quantitatively analyze the Gamergate controversy. We collected 1.6M tweets from 340k unique users using a generic methodology (which can also be used for other platforms and other case studies). Although focused on a narrow slice of time, we found that, in general, users tweeting about Gamergate appear to be Twitter savvy and quite engaged with the platform. They produce more tweets than random users, and have more friends and followers as well. Surprisingly, we observed that, while expressing more negative sentiment overall, these users only differed significantly from random users with respect to joy.
“Finally, we looked at account suspension, finding that Gamergate users are less likely to be suspended due to the inherent difficulties in detecting and combating online harassment activities.”
The report completely avoids the possibility that #GamerGate isn’t a harassment campaign, and instead relies on unproven and unsubstantiated claims that Twitter simply lacks the tools to detect “online harassment activities”. They also conveniently ignore that well over 90% of the million or so tweets in the Newsweek report were also neutral. The other percentages were either negative or positive.
It will be interesting to see how this data is used in media reports, assuming it’s used at all given that it vastly contradicts the notion that #GamerGate is an angry mob of irascible, uncontrollable, hostile harassers.
Ads (learn more about our advertising policies here)