,
8 tweets,
4 min read
Read on Twitter
You may have noticed that it seems like a lot of folks are claiming that Twitter is censoring or banning or silencing them. We decided to take a look to see which groups are feeling targeted and other things related to the phenomenon.
cc: @ZellaQuixote
cc: @ZellaQuixote
As a starting dataset, we used tweets containing "Twitter" and some form of the verbs "censor", "suspend", "ban", shut down", or "silence" from 1/1/2017 through 5/31/2019 (1871841 original tweets from 913237 accounts.) Who is hypothetically being censored?
Word cloud for the Twitter censorship tweets, with the search terms removed. Having seen a multitude of tweets from #MAGA accounts about being censored by Twitter, we're unsurprised by the prominence of "conservative" and "Trump," but they're not the only ones complaining. . .
This table shows the 30 non-trivial words that most often follow the censorship verb (censor/suspend/ban/shut down/silence) in "Twitter censors X" tweets. The right tops the list, but liberals, memes, women, antifa, and others have all been claimed as Twitter censorship targets.
We found a number of hashtags associated with claims of Twitter censorship, and correlated spikes in their use to events. Early spikes are reactions to Twitter policy changes, while later ones are based on user claims that X group is being censored (often without evidence.)
Finally, we compared the accounts tweeting about being censored by Twitter with a control group of accounts that have turned up in our projects over the last couple of years. We used a random sample of 6826 accounts from each set.
The results: based on average daily tweet volume, rate of duplicate text, frequency of tagging other accounts, and frequency of hashtag usage, the accounts complaining about being censored are on the whole somewhat spammier in behavior than the control group.
(separate thread on several botnets that attached themselves to the "Twitter censors X" traffic)
