Going Dark: Social Factors in Collective Action Against Platform Operators in the Reddit Blackout | MIT Center for Civic Media

Going Dark: Social Factors in Collective Action Against Platform Operators in the Reddit Blackout

Nathan Matias at CHI

This is a live-blog account of Nathan Matias' talk at CHI titled "Going Dark: Social Factors in Collective Action Against Platform Operators in the Reddit Blackout" which is work he carried out with Microsoft Research.



This paper describes how people who lead communities on online platforms join together in mass collective action to influence platform operators. I investigate this by analyzing a protest against the social news platform reddit by moderators of 2,278 subreddit communities in July 2015. These moderators collectively disabled their subreddits, preventing millions of readers from accessing major parts of reddit and convincing the company to negotiate over their demands. This paper offers a descriptive analysis of the protest, combining qualitative content analysis, interviews, and quantitative analysis with the population of 52,735 active subreddits. Through participatory hypotheses testing with moderators, this study reveals social factors including the grievances of moderators, relations with platform operators, relations among moderators, subreddit resources, subreddit isolation, and moderators' relations with their subreddits that can lead to participation in mass collective action against a platform.


Nathan mentions the choices that people have when they disagree with an institution. Who shapes the choices people have vis a vis institutions? Moderators have supported people online for more than 30 years. Now we have moderators and administrators on most online platforms. These moderators are at the centers of things when there is collective action against the platforms. He references the DailyKos Writer's Strike and Mechanical Turker Collective Action.


Moderators were appealing to the company for help and ways of addressing hate speech on the site. They had been talking about blackout for months. In July 2015, there were over 147,000 moderators on reddit. These people help people find jobs, seek mental help, celebrate cultural values and more. When the company let go a key employee who helped support moderators, one went dark, i.e. made their subreddit private. More than 2000 subreddits joined into this action, cutting off advertising revenue and public visitors to the site.


Also, there was a front stage and back stage thing happening. Reddit employees were calling moderators on back channels to try to negotiate.


Nathan is looking to link this to social movement theory. He's looking at Political Opportunity Theory which looks at who is going to take action at what time. There is another school of thought called Resource Mobilization Theory that looks at how institutions and people mobilize people to take action. That has been leveraged for online communities as well.


But, how do we study social action online during times of mistrust?

This has historical roots in Participant Action Research. Kurt Lewin was trying to understand behaviors of workers in a heavily surveilled situation.


Nathan started with data collection from the reddit API. He interviewed moderators. That fed into 12 explanations for blackout participation or non-participation. First, he offers qualitative findings and then builds those into a quantitative statistical model to see if those apply at scale. He then went back to participants and asked for feedback and advice in the form of participatory hypothesis tests. He grouped reasons into 5 themes - Grievances, Relationships, Resources, Group Isolation (how isolated the subreddit was from the rest of reddit), and Leader-community Relations. The statistical model predicted the likelihood of a subreddit to join the movement or not.


Of these five themes, there were two that had greater magnitude in predicting the likelihood of a subreddit to join the blackout. Grievances increased the likelihood and group isolation decreased the likelihood.


Nathan offers that his paper is also a methodological contribution because it provides a model for participatory hypothesis testing in which research hypotheses are vetted with their subjects.


Q: Victoria Bellotti from PARC. She asks about the availability of data on reddit vs other platforms where the data is closed. In some cases you have disagreements about religion or military actions. What kind of correspondences might you draw?


A: I also do work on online harassment. When people think about extremism they are trying to look at content and communication theory about how it spreads. There is value in looking at what are the conditions that lead people to be more likely to join a cause or movement. Hopefully this research encourages people to study more about people's wider social factors in different kinds of collective action.


Q: Dominic from UW. Did you come across rumors about what could have caused subreddits to go black but wasn't really a cause?


A: Yes there were a lot of rumors flying around across reddit. There was a belief that subreddits with employees were more likely to join that. But that was not the case. There were a number of cases where moderators held a vote. It's unclear how influential that was. They said they held votes to buy themselves time. Other cases moderators wanted to put it to a vote.

Q: Josh from MSU. Reddit still exists and is going thru transformation. Was it the event of the termination of this employee that caused this change? Or was the system graducally changing over time when it was ripe for this event? The termination then is a straw on the camel's back.

A: The work of predicting when a rare event will happen is challenging. In the qualitative work, many of the moderators who participated were long-term moderators. They had been asking for better tools and feeling pressure for years. They wanted to take some action to pressure the company. Moving forward there are follow-up studies. This could be seen as a model for a natural experiment. Are there peer influence hypotheses that could be tested? Those are a few that might be possible.