Political Bots, Subverting Twitter, and the Online Political Practices of Estonian Youth at AoIR16

Political Work Panel

This is a liveblog from the “Political Work” panel at AoIR16 on October 24, 2015 in Phoenix, AZ. This is not a transcript but recreation of people’s comments. Any errors are my own.

Architecture for Understanding the Automated Imaginary: A Working Qualitative Methodology for Research on Political Bots
Norah Abokhodair, Samuel Woolly, Philip Howard & David McDonald

This paper is led by Norah Abokhodair, is developing a working method for qualitative analyzing political bots. Summarized here: http://politicalbots.org/?p=314. Their research question: How are bots being used for political purposes?

They started with a set of definitions:

  • Bot = a software program that automates ‘human’ tasks on the web
  • Political bot = social bots, engage with human users. They mainly function on social media and are used to further specific political causes (for good, ill, or in-between)

The project has a three part research process: 1) comparative event data set, 2) international fieldwork with bot coders, and 3) computational theory building. The international field work involves interviews with people who build bots and track bots as well. We’ve looked into government contractors that track bots to combat activism online.

This paper focuses on stage one of the research: building the comparative event data set. They are documenting cases of political bot usage. They gather all media coverage of bot use around the world, and then use multi-coder content analysis of the media reports. They started in Hungary with students at Central European University, and triple coded all the media. They developed a Google Form that the coders would follow when coding each course.

The output of this is the contextual understandings of 100+ unique cases of political bot usage across 40+ countries. They noticed that anytime there was a political crisis or election there was use of political bots to manipulate public opinion. 

What is unusual about this research is that they are coding bots versus coding humans. Bots aren’t human but they are ‘technical social’ (Neff et al., 2012). Often this leads to confusion in coding about how to conceptualize the bots as political actors or on behalf as other actors. They find that it can be both. They are complementing with big (bot) data and ethnography of information to enrich the analysis.

Ultimately, they are working toward a multi-method and multi-disciplinary understandings. This follows the reconsidering without reinventing model (Markham and Baym, 2008). This is using hybrid methods form computers science and communication.

The paradox they are finding is: What happens when human agency is not necessarily the driving force of sociality of political change?

Their questions going forward include: How will political bots evolve,? How will they change particular policy and situated politics? What happens when the outcome of a piece of bot code is unknown? How can scholars move towards ethnography of information?

Samuel shared a anecdote about a bot coder in Amsterdam, whose bot started threatening people online and police brought him in for questioning because of the actions of his bot. Their is this interesting intersection with the law here and they are working with Ryan Calo to think through these questions.

Check out their website: www.politicalbots.org. You can nominate a political bot!

Subverting Social Media
sava saheli singh

Sava examines how academics use Twitter and what role that plays in our lives and scholarship. Twitter comes from a place of thinking that it’s an open space, but that ignores the way marginalized groups experience Twitter.

There have been some critics lately about how Twitter has been changed, they have nostalgia from their privileged early adoption. So how are features evolving not only technically but in how they are used and perceived by users?

She is focusing here on web use of Twitter, because mobile and app usage of Twitter requires a different set of assumptions.

Subtweets are used to create buzz and indicate to others something about a known person. This is like “plausible deniability” (Marwick and boyd). Subtweeting can take on a special quality when it comes from a private account. You can sometimes only see half of a conversation based on whose private accounts you have access to.

Block-unblock is a way to control your feed. There is no official way to force an unfollow. If someone is followed by someone they don’t want to be followed by they can block them and unblock them which unfollows them without that user knowing. This seems counter to the idea of Twitter that everyone can follow anyone else. But it’s a way to clean up your account from bots and low-use users, and a way to control your publicity.

Deleting tweets is another way to subvert Twitter. Users delete tweets for multiple reasons. Users can control their long-term identity on Twitter, but creates problems where interlocutors look like they are talking to themselves when the other have of the conversation disappears.

In the case of Politwoops, Twitter shutt down that service’s access to the API because they tracked the deleted tweets and other behaviors, responding to their users wants just-in-time.

Promoted Tweets are mostly working on behalf of the celebrities and brands that use them without thought of how that interrupts flows of conversation on Twitter from recipients. These are like pop-up ads, users have to expend emotional labor to dismiss or ignore them. Trolls who sent out and promoted racist harassment showed how Promoted Tweets can be abused.

.@ is used to broadcast a discussion to your followers, but can also be used as an offensive bullying mechanism.

Storify is a useful way to preserve and present conversations. But it can also be used as a form of abuse, as curators selectively include and comment on a conversation.

Retweeting is a way to rebroadcast something to your followers. The earliest form of retweeting required RT or MT plus copy and pasting the text. The new retweet commenting allows you to use the full 140 characters while retweeting the cat. Users before had more flexibility about how they used it. Now it’s harder to shame Twitter users. Instead of a Tweeting a link to the original post, now users sometimes screenshot the original tweet to draw negative attention without alerting users.

Fav was originally meant to be for bookmarking a Tweet, now it is more broadly used as liking. Some users prefer faving to retweeting so that they don’t have explain context to their users. Twitter’s new feature “While you were away” breaks the ephemerality of Twitter and changes faving behavior social signal.

The most vulnerable often need to find ways to subvert the features of a platform like Twitter. Researchers need to keep up with both how features change and how users employ them. Justifications for feature changes may follow a commercial logic rather than a user logic. More and more the 140 character limit is being transformed based on what counts and what doesn’t. Ultimately, we need new literacies to keep up with all these changes.

Perceptions of Participation and the Share Button
Katrin Tiidenberg & Airi-Alina Allaste

This research looks at how Estonian youth use social media, how they perceive it, and whether it represents non-institutional political activity.

Conventional political participation in Europe has decreased since 1970s. Pessimistic scholars and media see this as a trend of disengagement. While optimistic scholars see other political activities like boycotting and buycotting on the rise.

The lived experience of youth is redefining political participation and vice-versa. Scholarship looks at phenomena like austerity politics, social fragmentation, and the individualization of the life-course. Tiidenberg and Allaste are following the discourse on new political repertoires that represent a kind of everyday, ordinary political participation.

There is a lack of consensus in scholarship about which uses of social media are political. Three mechanisms of social media seem to support political participation: 1) provide information, 2) introduce social pressure, and 3) enhance discussion among peers.

Tiidenberg and Allaste looked at sharing and liking, commenting, and signing petitions. This is part of the MYPLACE project 2011-2015 surveying 14 countries in Europe. The sample is 16-26 year olds. The field work was in late 2012 to early 2013. It included the MYPLACE surveys of 1251 in two locales in Estonia (both ethnic Estonia areas and Russian areas). And then did 60 interviews in those two locales.

Note: Activism has a bad connotation in Estonia because it is tied to Russian communist fake activism. You don’t want to be labeled an ‘activist’ in Estonia.

Only 20% of informants believe politicians work in the interest of citizens. In the Russian speaking area 40% saw themselves as inactive, but their own definition only covered traditional examples of political.

21% Estonian-speaking and 8% Russian-speaking youth signed petitions. These were for for a range of issues like education and environment. Interviewees said they spent a lot of time deciding what it meant for them to sign a petition. They might choose to only sign things others wouldn’t judge them—they were political safe like the environment. They were doing what might be read as impression management.

Commenting was used by some informants to situate their own opinions in the larger discourse. Others preferred to read comments to be informed but didn’t want to participate out of fear of conflict and a broader rational choice justification saying that there is no point because I have the same point to make or they have a completely different opinion from others.

Sharing practices were dominated by efforts to help others: notices from police or awareness of broader issues. But the only political content they shared was humor. This could be seen as a way to avoid conflict by neutralizing the political content. They made decisions to not share out of impression management again. But the importance of the issue was a factor as to whether others would want to see it. Liking was seen as less approval than sharing but still participating. Everyone liked things political but not everyone shared them.

The perceptions of effectiveness of social political activity was low. Young Estonians express interest in keeping informed about things considered political. But the social imaginaries among youth indicated perceptions that 1) contributing on social media makes no difference, 2) politicians don’t care about young people, and 3) the things i do are not political.

Optimistically, we can interpret Estonian social media political practices as finding ways to engage in everyday spaces. But we can also read this as trends in self-censorship and a “civics without politics.”

We need more research about why young citizens have a low sense of efficacy about practices and self-censor their participation. We also should know more about how the perceptions of social media themselves change their perceptions of political participation given media representations about fear over abuse from sharing.