Reading:
[1] An Army of Me: Sockpuppets in Online Discussion Communities
Summary:
The authors of this paper have devoted their energy towards talking about sockpuppets in online discussion communities. So as to comprehensively study sockpuppets, and their associated online behavior the authors obtained data from nine different online discussion communities consisting of 2,129,355 discussions , 2,897,847 users, and 62,744,175 posts. They then decided to identify sockpuppets by using a combination of three elements: their IP address, their activity in the discussion post, and the time at which the comment(s) were made. By using this combination of factors, they were able to formally define sockpuppets — “a user account that post from the same IP address in the same discussion in close temporal proximity at least 3 times.” Utilizing this formal definition and an analytical model, the authors deduced 1,623 sockpuppet groups and 3,656 sockpuppets from nine different online discussion communities. Outcome of the project ensued in plethora of intuitive but interesting results including but not limited to the following list:
- Sockpupptets start fewer discussions, and post more in existing discussions.
- Sockpuppets tend to participate in discussions with more controversial topics.
- Sockpuppets are treated harshly by the community.
- Sockpuppets in a pair interact with each other more. etc.
Reflection and Questions:
I had really never thought about this area of research and hence this reading ensnared my attention and interest. Howbeit, as I read through the paper it seemed as if it was more focused towards the pretenders and less on the non-pretenders and that reflected in the way they defined sockpuppets — which is totally fine but according to me the authors should have mentioned their focus somewhere in the introduction. Since I didn’t find quality material for non-pretenders, I started thinking how would I define sockpuppets with respect to non-pretenders? Assuming complete access to a users profile, I would start by correlating basic information of the user; for instance their birthday, secret questions, name (in some cases), small variations in username, family information (if available), and contact information. Since non-pretenders do not masquerade, simply use different accounts for different use cases, I would assume that they would not have any reason to manipulate their basic information — unless the platform prevents the user from doing so. Whilst reading the paper, I started to contemplate what could be the embolding factor behind puppetmasters? The only reason I could think was the motivating factor to push their/their sponsors’ political, and ideological agenda, or dilute the opponents agenda. Howbeit in both cases I would assume that puppetmasters would be more articulate in their writing to effectively sway the audience in/against either direction, but the results of this paper — that sockpuppets write shorter sentences with more swear words and use more personal pronouns — were counterintuitive to me. As I was reading through the fifth section of the paper, it occurred to me to think about how long these accounts have been active ? or how frequently does a supposed puppetmaster create new accounts? I am not sure yet what new things we might discover by seeking answers to these questions, but I think these are interesting questions. Another correlation that I strongly thought about was to check if sockpuppets are recycled among different puppetmasters/groups? If we find this to be true, and do some analysis on the topics that these sockpuppets try to propagate or demolish the support of, then we can group the groups according to their affiliations; and if we add a spatial aspect to the groups of groups then we may be able to associate and identify what kind of ideologies are wide-spread in what part of the world. We might also be able to find out if a group is trying to propagate it’s own ideology or demolish another regions’. For instance a group from country X is spreading hate towards topic Y, but as a matter of fact topic Y is appreciated in this country X, then we know that this group is demolishing ideology in a different region and so is true for the opposite where topic Y is hated in country X.
