Clean Up The Internet wrote to Twitter two weeks ago, regarding claims made by the company about the nature of racist Tweets directed at members of the England men's football team after the European Championship Final. We requested that Twitter provide more explanation as to how it had arrived at its conclusions, and make available the evidence on which it had based them.
Two weeks later, Twitter has failed to reply, so we're making the letter public. This is just the latest example of a large social media company issuing claims about what "the data" says about the safety of its platform, which it then fails to substantiate, and refuses to open up to independent scrutiny.
The draft Online Safety Bill includes provisions to require social media platforms to produce transparency reports, and for improving access to data by independent researchers. This latest episode illustrates how important it is that these measures are strong enough.
17 Aug 2021, 14:35 Dear Mr Dorsey, (cc Nick Pickles, Katie Minshall, and Sanjay Bhandari (Kick It Out)) I write with regard to Twitter’s blog post dated 10 August, “Combatting online racist abuse: an update following the Euros” (https://blog.twitter.com/en_gb/topics/company/2020/combatting-online-racist-abuse-an-update-following-the-euros). Clean Up The Internet welcomes your condemnation of “racism in all its forms”, and of the abuse directed at England football players, using your platform, following the Euro 2020 Final. We also welcome your stated desire to “contribute to the shared understanding of these behaviours online” and to become a “diverse, inclusive, and accessible tech company”. In light of the subject matter, I am copying in our friends at Kick It Out, with whom we have been sharing our concerns. Your blog post presents a series of claims about the authors of the racist Tweets, and uses these claims to draw conclusions about what would (and wouldn’t) help reduce racist abuse in the future. I’m therefore writing to you to request additional information, which would enable us to better understand how you arrived at these claims and assess their accuracy. Firstly, please can you share the data upon which you based this analysis? In particular, please can you make available details of the 1622 Tweets you say you removed “during the Final and in the 24 hours that followed”, and provide details of the accounts which you say you have suspended such as the account handles, bios and profile pics? Where you are unwilling or unable to provide this data, please could you explain your reasoning for this given that this was all data which would have been in the public domain up until the point when Twitter took the decision to remove it. Secondly, we have some specific questions about some of the conclusions which you claim this analysis has led to. Your claim that “99% of the accounts suspended were not anonymous” · Please can you explain how Twitter defines “not anonymous”? Can you confirm whether or not Twitter believes it knows the real names and location of the owners of these suspended accounts, and how you know that. · Please confirm how many accounts you suspended. Did you identify accounts posting racist comments that you did not suspend? · Please can you explain what data you relied on, and what methodology you used, to “identify” the owners of these suspended accounts? · Please can you confirm whether or not you have tested the accuracy of your identity assessments and if so how? · Please can you confirm whether or not the owners of these suspended accounts would be readily identifiable (or “not anonymous”) to: -the targets of their abuse? -ordinary twitter users? -law enforcement agencies? · Please can you confirm in how many instances you have passed the details of these “not anonymous” accounts to law enforcement or football agencies? Have you informed other social media companies so that they might use this information to check against their own users? · Where the account holders were identifiable, have you been able to prevent them setting up new accounts and if so how have you done this? Your claim that “ID verification would have been unlikely to prevent the abuse from happening” · Please can you explain your reasoning for this conclusion? · Please can you explain how you track multiple account ownership and with what level of confidence you are able to identify where an owner of a suspended account is also linked to other accounts? · Please can you provide details of how many of the owners of suspended accounts have in the past operated, or currently operate, accounts other than those from which the abusive Tweets were sent? · Please can you provide details of how many of the authors of abusive Tweets who have also been linked to other accounts have previously had enforcement action up to and including suspension against other accounts? · Please can you provide details of how many owners of suspended accounts have subsequently created new accounts, or have attempted to do so but been prevented by Twitter? Your claim that the “UK was - by far - the largest country of origin for the abusive Tweets we removed” · Please can you explain how you assessed the country of origin for abusive Tweets? · How would you explain the discrepancy between your assessment of the balance of UK/non-UK accounts and that of the UK Football Policing Unit? They stated on 5 August that of the 207 posts following the Euros Final, across all social media platforms, which they had identified as meeting a criminal threshold, 123 accounts belonged to individuals outside the UK. How many of the posts referred to by the UK Football Policing Unit were on Twitter, and what was the balance of UK/non-UK accounts for the Twitter posts? · Have you conducted any comparable assessments of the balance of UK/non-UK accounts sending racist abuse to footballers, beyond those associated with the Euros final? For example do you have any assessments of the origins of racist tweets directed at Premier League footballers over the course of the last season, similar to the study of Twitter accounts conducted by the PFA and Signify? I’m sure you would accept that independent scrutiny of your claims is essential if they are to enjoy trust and credibility, and so we look forward to your prompt reply. Yours Sincerely, Stephen Kinsella Founder, Clean Up The Internet