Onlinecensorship.org is pleased to share our first report "Unfriending Censorship: Insights from four months of crowdsourced data on social media censorship." The report draws on data gathered directly from users between November 2015 and March 2016.
We asked users to send us reports when they had their content or accounts taken down on six social media platforms: Facebook, Flickr, Google+, Instagram, Twitter, and YouTube. We have aggregated and analyzed the collected data across geography, platform, content type, and issue areas to highlight trends in social media censorship. All the information presented here is anonymized, with the exception of case study examples we obtained with prior approval by the user.
Here are some of the highlights:
This report covers 161 submissions from 26 countries, regarding content in eleven languages.
Facebook was the most frequently reported platform, and account suspensions were the most reported content type.
Nudity and false identity were the most frequent reasons given to users for the removal of their content.
Appeals seem to present a particular challenge. A majority of users (53%) did not appeal the takedown of their content, 50% of whom said they didn’t know how and 41.9% of whom said they didn’t expect a response. In only four cases was content restored, while in 50 the user didn’t get a response.
We received widespread reports that flagging is being used for censorship: 61.6% believed this was the cause of the content takedown.
While we introduced some measures to help us verify reports (such as giving respondents the opportunity to send us screenshots that support their claims), we did not work with the companies to obtain this data and thus cannot claim it is representative of all content takedowns or user experiences. Instead, it shows how a subset of the millions of social media users feel about how their content takedowns were handled, and the impact it has had on their lives.
The full report is available for download and distribution under Creative Commons licensing.