by Tom Chivers
Wednesday, 11
August 2021

Do racist Euros tweets tell us anything about the UK?

The information provided by Twitter is meaningless
by Tom Chivers
Marcus Rashford after he missed his penalty (Photo by Visionhaus/Getty Images)

Last month, in the wake of the Euro 2020 final the three players who missed their penalties — Bukayo Saka, Marcus Rashford and Jaydon Sancho, all young black men — were subjected to horrible racist abuse online. 

One of the few mitigating factors there was that, according to Reuters, 70% of that abuse came from accounts registered overseas: that is, it wasn’t, mainly, British people being racist.

But now Twitter has published a blog post saying that “The UK was by far the largest country of origin for the abusive Tweets we removed on the night of the Final and in the days that followed.” So was that 70% claim wrong?

Well: possibly not. Imagine that 100 racist tweets were sent. (There were more than that, of course.) Now imagine that 70 of them were sent from 70 different countries: one from Afghanistan, one from Albania, one from Algeria, all the way down to Iraq. And then 30 of them were sent from the United Kingdom.

Now, 70% of racist tweets were sent from overseas: but the UK is by far, by a factor of 30, “the largest country of origin”. Given that Twitter didn’t say that the UK was the source of the majority of the abusive tweets (despite some bad reporting claiming otherwise), it seems likely that something like this is the case.

Perhaps a more interesting question, though, is: does this matter? Does it tell us anything? You can easily find hundreds of people saying that it shows that this is a racist country. But I think it tells us nothing at all: not that the UK is racist or that it isn’t. It is completely uninformative.

A total of 1,622 tweets were deleted, according to Twitter. That is roughly one-quarter of one second’s worth of tweets on the site. Ten committed racists (or edgelord dickheads) could send that many tweets in half an hour without being able to touch-type. Even if they were all from the UK, and all from different users, we’d hardly be surprised to learn that there are 1,622 racist idiots among the 16 million UK Twitter users: that’s 0.01%, and I’m sure there are a lot more racists than that. So it doesn’t tell us that the UK is racist. 

And more than that: the 1,622 tweets will almost certainly not have been all the “racist” tweets. Sunder Katwala of British Future has been doing a sterling job over the last few years establishing what Twitter will delete. He notes that, for instance, “No blacks in the England team – keep our team white” and “Marcus Rashford isn’t English – blacks can’t be English” are not considered “in violation” of Twitter’s rules.

So you can’t say “there were only 1,622 tweets, Britain isn’t racist” either. All you can say is “according to Twitter, 1,622 tweets broke their arcane rules, and some unknown percentage of them came from Britain”. This information tells us, I think, almost literally nothing at all.

But what can we say? We can say that, firstly, British social attitudes have changed, and that far fewer people hold racist beliefs than they did 30 or 40 years ago; but also, Katwala and others have found that their daily experience of racism has gone up.

Katwala thinks (and I agree) that social media is the conduit there. As I said above: 10 racists could send a thousand racist tweets in an hour. And they could easily find enough people of colour to send them to. A growing number of people not being racist doesn’t greatly help, if it is ever easier for the declining number of actual racists to find targets for their racism.

This does suggest that there are actual things we can do, though. Obviously encouraging a reduction of racism in society is good in its own right, but it won’t help much with online abuse. Instead, it’s a moderation problem. A large fraction of racist online abuse seems to come from a small group of Twitter users who get banned and simply make new accounts, hundreds of times. If Twitter were to stop letting them come back on each time under a thinly altered new identity, that would be a good start.

If we want to reduce the experience of online racism, then a really good way to start would be for Twitter (and other social media sites) to stop saying “there is no place for racism on our platform” and start paying for moderators. It would have a much more direct and immediate effect than any number of “Kick It Out” campaigns aimed at a wider society which is, on the whole, much less racist than it used to be.

And if you want a more heartening story: Katwala thinks the total number of anti-racist tweets after the final outnumbered the racist ones by at least 99 to 1. Most people, on Twitter and elsewhere, are not dreadful.

Join the discussion

  • There’s another pretty obvious point to be made here.
    Reuters said:

    70% of that abuse came from accounts registered overseas

    Whilst Twitter said (my emphasis):

    The UK was “by far the largest country of origin for the abusive Tweets we removed on the night of the Final and in the days that followed

    We don’t know what proportion of the total online abuse was removed so these numbers could be wildly different from the start.

  •  Most people, on Twitter and elsewhere, are not dreadful

    I am finally beginning to get what Twitter (and other social media) do. Each variety of social media is different but there are sets of mechanisms in common.
    Let’s take one step back. Advertising (to sell products, or religion, or values, or anything at all, works by subliminally creating patterns of associations that have emotional force to individuals who want to participate in the message. It doesn’t affect people who are not ready to participate – there is a principle of tacit consent. Everyone has multiple, overlaid models of complex social learning they use to filter or allow through what they process. What advertising does, is it creates forms that hit enough triggers to cause an engagement with the message – potentially against the conscious wishes of the recipient. This is essentially a creative activity, and inherently bespoke in nature – a good advert is as good as a good piece of art. But this entire model is blown out of the water by a different approach.
    Algorithmic technologies don’t need to go for this approach because they can scale in a way individual humans never can. A high speed, high volume scatter-gun generates a combinatoric volume of trigger possibilities that in effect relies on most of the flow to be filtered out at an individual level, but some combination or other of the message form will get through and cause engagement – and then other innate human mechanisms take over and you are hooked. For example, Tom Chivers claims “…10 racists could send a thousand racist tweets in an hour…”. But any coder worth the salt would find such an approach hopelessly inefficient – what do you think bots are for? If you can create code to make your racist (or anything else) tweets, why on earth would you need actual racists? You just want one or two racists to teach the code models what to do. This is what I mean by scale. And since algorithms are not human scale, they can replicate, magnify and transmit simultaneously to all corners at speeds that will overwhelm any bespoke human process.
    The useful way to view Twitter is not as people. You are people. I am people. Twitter is a petri-dish of biological and social responses – a bunch of randomers you cannot look in the eye and you have no visibility of. Don’t get trapped, don’t engage with Twitter as just a user. Unless that is, you are doing so for very good reasons, like making money. Be the pusher, not the junkie. Learn to code.

  • To get involved in the discussion and stay up to date, become a registered user.

    It's simple, quick and free.

    Sign me up