Monday, October 21, 2019

Mapping Hate: A Social Network Analysis of the Gab Social Network


Context:
In recent years alternate social network sites without the more rigorous content moderation restrictions of more mainstream sites have proliferated across the internet. These sites have arisen for a variety of reasons, from pure financial profit to a desire to create distinct online ideological communities. Often, sites with more liberaterian outlooks on moderation and free speech have been accused of harboring radical far-right elements that extort racial supremacist and hateful views far outside the mainstream. One such site is Gab, a “twitter clone” that has been cited for its far-right user base and willingness to harbor neo-Nazi, White Supremacist, and other hate based groups[i]. This is supported by a recent content analysis that noted that Gab users utilize hate speech at a far higher rate than more mainstream sites such as Twitter, a fact that was brought into stark view when the Pittsburgh Synagogue Shooter posted a hate-filled screed on the site before his October 2018 terror attach. Previous analyses of Gab have focused much more heavily on content analysis, while past SNA’s of right-wing groups have focused much more on organizational relationships between varying groups[ii].

Research Question:
This social network analysis will seek to identify the evolution of alt-right messengers on the Gab Social Network, and identify how they have changed as the movement has gained prominence. We will conduct an SNA at three separate instances of time surrounding major right-wing terror attacks, and identify the changing network characteristics of groups discussing these events in a positive light. Possible events include the 2017 Unite the Right Rally in Charlottesville, 2017 Portland Train Attack, and/or the October 2016 arrest of the Garden City bomb plotters.

Data Sources and Analysis:
The dataset for the 2018 “What is Gab?” analysis is an excellent existing publicly available data source. Researchers utilized Gab’s API to crawl the social network and obtain data for the most popular users utilizing a snowball sampling methodology. The researchers collected data on the most popular users and their posts, the followers of those users, and their followings. This should allow us to construct a dataset of the most influential members of Gab, as well as members of their immediate network.

While it may involve some form of content analysis, we can identify the most up-voted and shared posts surrounding the selected events. We can then visualize those networks to see which members of the network are the most influential and responsible for disseminating the messages most widely. Given that the dataset is constructed from the most popular accounts on the network, it stands to reason that the originators of posts would have high degree measures, other centrality measures for other parts of the network such as eigenvector could present a more nuanced and detailed story.

What this SNA could do:
This SNA could provide an interesting snapshot of a growing social media network and messaging strategy of a growing online community that has only grown as a concern for policymakers. Identifying the shifting leadership structure of those responsible for the majority of hateful messaging on the network can allow us to understand where this movement is going and how it is changing.

1 comment:

Christopher Tunnard said...

A good idea, but yor overall research question needs to be more meaningful and precise. For instance, you need to think a bit more about what you mean by "evolution of alt-right messengers" and "changing network characteristics." You also need to consider more deeply which network measures will be the most significant. For instance, how do you define popular? Or put a value on an up-vote or a share? And given the biased samples (most popular accounts) what if anyimpact will this have on your conclusions?

All to be teased out during the course. I look forward to seeing it develop.