In the wake of the assault on the US Capitol in January, Twitter took decisive action against false claims of election fraud by suspending 70,000 accounts associated with the QAnon conspiracy.
However, new data shows that almost all of the most prolific spreaders of electoral fraud misinformation were within the orbit of the Republican mainstream, and were largely left to their own devices by a Twitter purge that appeared to focus on fringe conspiracists.
This comes as Parler, the “pro-Trump” alternative to Twitter, has gone live once more after the social network closed for more than a month. Its tech suppliers had withdrawn their support over fears the app was used to incite violence before the storming of the Capitol.
But our analysis shows US voters searching for baseless claims of electoral fraud did not need to look further than Twitter: the social network was much less likely to suspend accounts that had shared bogus claims if these accounts hadn’t also promoted QAnon-related hashtags.
Researchers at Cornell University collected nearly four million tweets relating to voter fraud which were posted in the last three months of 2020. Users were then automatically graded by their influence in spreading the conspiracy, and clustered into groups based on their interactions with other accounts.
The team found that while 40 per cent of users promoting the voter fraud claims fell into a cluster linked to the QAnon conspiracy movement, 52 per cent belonged to a community much closer to the Republican Party mainstream.
More than one in ten users in the fringe group had shared QAnon-related hashtags – more than eight times the rate of the mainstream group. The New Statesman found these fringe group users were also 247 per cent more likely to share explicitly QAnon-related YouTube videos than users in the more mainstream group.
Of the 53 elected politicians, all Republican, who were among the 10,000 most influential proponents of the election fraud conspiracy, all but three were in the mainstream cluster (or 94 per cent). This group was also much more likely to share links to conservative or Republican-supporting news outlets such as Fox News, the Texan and the Houston Chronicle.
By contrast, users in the QAnon-linked group preferred non-traditional news sources such Gateway Pundit, DC Patriot and the Epoch Times.
Twitter’s purge largely overlooked the more mainstream group, appearing instead to focus on the QAnon-linked cluster, whose members made up 71 per cent of the accounts suspended.
Yet, data suggests the more mainstream users had greater influence over the spread of the claims of voter fraud.
Of the 100 most influential promoters of the claims, just one belonged to the QAnon-linked cluster. The other 99 belonged to the mainstream group, yet only 12 of these were suspended.
On the whole, Republican politicians were spared suspensions. The New Statesman found that, of the 53 senior elected officials who were among the 10,000 most influential promoters of the voter fraud claims, Donald Trump was the only one to be permanently suspended.
Twitter declined to comment on the findings, or to provide details of how users were targeted for suspension.
“The fundamental problem is that we don’t have transparency,” says Sam Gregory from the human rights group Witness.
“We have transparency on what happened because, unlike other platforms, Twitter gives more API access. But we don’t have a huge amount of transparency on Twitter’s decision-making, what right to appeal people had, and how consistently the suspensions were applied.”
The research team at Cornell calculated the influence of users based on their activity, with an influence score representing their ability to spread information to other users in the network.
Press Gazette: UK competition authority chief says “compelling case” to regulate most powerful tech platforms Part of New Statesman Media Group
While members of the QAnon-linked cluster were more prolific in their online activity, more than two-thirds of their retweets and quote tweets (69 per cent) were directed at amplifying the tweets of users in the mainstream group.
User clusters were automatically extracted from the data using a community detection algorithm.
“The algorithm tries to find places in the network that are more interconnected than other places in the network,” says Mor Naaman, who led the team of students that carried out the study.
“What we see in the data is that, within the broader community of people who are promoting those claims, there’s a sub-community of people that tend to interact more with each other and that are much more likely to be associated with QAnon than the other communities.”
Investment Monitor: Big Tech braces for EU regulatory assault Part of New Statesman Media Group
Human rights organisations are campaigning for social media companies to provide greater transparency and ensure greater consistency in their moderation policies.
“Not only do we not know how they picked this group in the US, we don’t see this being consistently done in other locations in a way that addresses direct harm,” Gregory says. “Although we may not feel sorry for the QAnon users, the other side of this is that human rights defenders and marginalised people are constantly being pushed off platforms in ways which are completely unaccountable.”
Tech Monitor: The bills tech giants targeted in 2020 Part of New Statesman Media Group
The government in India has demanded that Twitter suspend accounts linked to the country’s ongoing farmers’ protests.
“With the farmers’ revolts in India right now, it ties into a lot of the issues about who we consider terrorists and extremists,” Gregory says. “This has been an ongoing issue for all these platforms.”