This Account Doesn’t Exist: Tweet Decay and the Politics of Deletion in the Brexit Debate

Abstract

Literature on influence operations has identified metrics that are indicative of social media manipulation, but few studies have explored the lifecycle of low-quality information. We contribute to this literature by reconstructing nearly 3 million messages posted by 1 million users in the last days of the Brexit referendum campaign. While previous studies have found that on average only 4% of tweets disappear, we found that 33% of the tweets leading up to the referendum vote are no longer available. Only about half of the most active accounts that tweeted the referendum continue to operate publicly, and 20% of all accounts are no longer active. We tested whether partisan content was more likely to disappear and found more messages from the Leave campaign that disappeared than the entire universe of tweets affiliated with the Remain campaign. We compare these results with an assorted set of 45 hashtags posted in the same period and find that political campaigns present much higher ratios of user and tweet decay. These results are validated by inspecting 2 million Brexit-related tweets posted over a period of nearly 4 years. The article concludes with an overview of these findings and recommendations for future research.

Facing the volatility of tweets in altmetric research

“…About 3.5% of the tracked scholarly tweets became unavailable at some point in a period of 56 days. This is a small percentage, but it may be a nontrivial amount when working with older and larger sets of tweets, or tweets related to controversial topics or accounts whose deletion or suspension 8 would lead to a cascade of related Twitter data becoming unavailable (Bastos, 2021; Fang et al, 2020).…”

“…In 2016, Zubiaga (2018) rechecked the completeness of 30 Twitter datasets associated with different real‐world events posted between 2012 and 2016, finding that, on the whole, 18.6% of the tweets were unavailable during the data rechecking. Bastos (2021) reported that 33% of the nearly 3 million tweets about the Brexit referendum posted in 2016 were no longer available 3 years after the vote. In addition, by rechecking the 42.5 million scholarly tweets recorded by http://altmetric.com until October 2017, Fang et al (2020) found that 13% of the tweets had become unavailable by September 2019.…”

We love to hate George Soros: A cross-platform analysis of the Globalism conspiracy theory campaign in Brazil

“…Six of these 10 most shared links were unavailable at the time of the analysis. Given the opaqueness and the politics of deletion implemented by these platforms, relying on social media data as a record for public deliberations can indeed be difficult (Bastos, 2021). Notwithstanding, not all types of problematic content are removed by the platforms and not only fake content and disinformation are deleted.…”

Editorial: Five challenges in detection and mitigation of disinformation on social media

“…Perhaps more worryingly, there are more messages associated with the Leave campaign that disappeared than the entire universe of tweets affiliated with the Remain campaign (Bastos, 2021c). In fact, the list of hashtags tweeted over one thousand times with a deletion rate of 40% or higher is largely restricted to Leave terms: voteleave , votein , leaveeu , ivoted , voteout , beleave , cameron , inorout , ukip and eng .…”

“…Lastly, even if the rapid turnover and short shelf life of social media content constitute an expected affordance of social media communication, it is hardly a desirable design of political communication and deliberation across social platforms that may be further subjected to Editorial artificial manipulation and false amplification. In other words, while the short shelf life of social media posts may be a reasonable expectation, as is the expectation that one should have control over its own personal data (Ausloos, 2012), this poses considerable challenges for deliberation and informed public debate around matters where the issue being deliberated on is constantly disappearing from public scrutiny (Bastos, 2021c).…”