Behind the hashtag: Who’s tweeting about #SurrenderAct ?

If you’ve been following the latest news about Brexit, then you’ve probably heard about the so-called ‘Surrender Act’.

It’s Boris Johnson’s way of describing the Benn Act, passed by Parliament earlier this month to prevent No-Deal Brexit. This compels Johnson to seek an extension to Article 50 if he can’t reach a deal with the EU by October 19, 2019.

Johnson’s supporters didn’t approve of this legislation. They claimed that the Act would ‘undermine’ Britain’s negotiating power with the EU.

#SurrenderAct immediately started trending on Twitter. But who exactly was tweeting it? I jumped into the analytics to find out.

When did the hashtag start?

When analysing a hashtag, I usually begin by checking when it was first tweeted, and by whom. #SurrenderAct was first used by an account that really didn’t want to look like a bot…

Below we see a sharp spike in activity around the hashtag. It was tweeted over 3000 times over 12 hours (mainly during the UK night time).

So who else is tweeting about #SurrenderAct? Below are the top 10 most active hashtag users. In the rest of this post, I’ll put these accounts under the microscope.

Bot, cyborg, or organic human?

You’re probably wondering how many of these accounts are bots. Time for a quick reminder about what bots can (and can’t) do on Twitter. They’re pieces of code designed to amplify a particular hashtag, user or keyword. DFR Lab has a useful guide for spotting automated accounts.

The most obvious indicator of ‘bot-ness’ is high levels of activity, i.e. non-human tweeting patterns. Other top indicators are anonymity: e.g. no photo, or a generic one, a non-specific (usually political) bio, and a vague location, e.g. ‘England’, and amplification: only retweeting or liking other people’s tweets – i.e. boosting their messages in a quick and low-effort way.

Bots are less effective in human-to-human engagement, such as arguing with other Twitter users. That’s more likely to be human operators (or cyborgs, which mix bots with humans).

So, if boosting #SurrenderAct was the main purpose of these accounts, then we’d expect to find evidence of typical bot-like behaviours.

Let’s take a look at three interesting accounts within the top 10.

1. The Hyper-Prolific Tweeter

This account is new to Twitter, having joined in March this year. It has no photo (only the typical ‘egg’) and no bio. Definitely low effort.

But its rate of tweeting is impressive! During a short space of time, ‘christine’ has achieved a rate of over 1000 tweets per day.

Researchers cite a number of different benchmarks for identifying ‘bot-ness’. The Oxford Internet Institute says it’s an average of 50 tweets per day. DFR Lab is more generous. It claims that 72 tweets per day would be suspicious, and over 144 would be ‘highly suspicious’.

Remember too, that retweeting is faster and lower effort than creating replies or original tweets.

As shown above, ‘christine’ is going full bot. 100% of the account’s activity is retweets, all from the Twitter for iPhone app.

2. The Latent Islamophobe

‘Sue Reap’ is at number eight among those who most tweeted #SurrenderAct. There’s some interesting things going on with this account. Its bio is peppered with Tommy Robinson references and hashtags.

The account joined Twitter over seven years ago. But a couple of quick advanced searches shows that it didn’t tweet anything for most of 2012 or 2013.

Or, perhaps it did, but those tweets got deleted…It’s not easy to know.

Suddenly, ‘Susan’ springs into action in late 2013/early 2014 with a flurry of anti-Muslim tweets.

We can see that this account has a suspiciously high activity rate, producing 126.88 tweets per day, of which 22% is replies.

This rate puts the account close to the DFR Lab’s ‘highly suspicious’ bracket of 144 tweets per day.

So has ‘Susan’ given up on Tommy?

Not in the slightest. He’s still foremost in her mind, right up there with leaving the EU. It’s practically an obsession.

3. The ‘true-blue’ Brexiteer

This account is likely to be ‘organic’, i.e. a normal human user. It’s become quite Brexity in recent years, but still within the realms of normal human behaviour.

‘Pat’ was an early adopter of Twitter, joining in 2009, possibly when he/she was 55 (guessing from the handle). That would put them in their mid-60s now; the typical Brexit voter demographic.

At the beginning, ‘Pat’ tweeted everyday comments about garden parties and Michael Jackson. There was no sign of anything political.

In April 2016, when the referendum had been announced, ‘Pat’ was tweeting happily about normal things: celebrities, photography and TV shows.

But come May, as Britain inched closer to the date of the referendum, Pat’s political side suddenly became apparent. Out came the pro-Brexit tweets.

Despite this, the account is still within the realms of being normal. An activity rate of 33 tweets per day is nowhere near ‘botness’. What’s more, the 82% of replies shows that this account engages a lot with other users, rather than simply retweeting things blindly. This is not typical ‘bot’ behaviour.

It’s likely to be a typical older Brexit voter who has become somewhat radicalised by the debate’s tribal nature (it’s not just Brexit voters; but happens to both sides).

These accounts form just a tiny sample of the millions of accounts out there engaging with political content.

Key takeaway: Don’t just assume everyone is a bot; instead think critically before jumping to conclusions.

A tale of two PMs: Facebook, astroturfing, and social proof

There’s something odd about the Prime Minister’s Facebook page.

Underneath every post, especially those about Brexit, are hundreds of responses. This isn’t unusual for the page of a public figure, but the style of the responses didn’t ring true.

They are all very similar; short utterances of praise for Boris Johnson, repeating words and phrases such as ‘brilliant’, ‘fantastic’, and ‘support Boris 100%’. Each comment is festooned with Facebook’s emojis, mainly representing positive sentiments of ‘like’, ‘love’ and ‘laugh’.

This behaviour feels odd. I’m not denying that a lot of genuine people do support Johnson, but it’s suspicious for so many to consistently comment on his posts in this distinctive and repetitive fashion.

Screenshot of Boris Johnson’s Facebook page, with a selection of comments about Brexit.

Let’s contrast this with the page of his predecessor, Theresa May, specifically her Brexit-related posts. Here we see a very different scenario.

Responses to May’s posts tend to be a lot more varied, in content, tone and length. Some disagree with her. Others support her. But most are expressed in more depth and sophistication of language than the short repetitive replies on Johnson’s.

In short, the responses on May’s page look far more likely to be ‘organic’ (i.e. produced by real people behaving naturally) than the majority of those on Johnson’s. It’s possible that Johnson’s page is using artificial amplification techniques, which may include fake followers.

Screenshot of Theresa May’s Facebook page showing a sample of typical comments about Brexit. Note the contrast with Johnson’s page.

Facebook locks its data down tight, so it’s hard to run further analysis to determine for certain whether the Johnson supporters are part of an organised campaign.

But we can draw from previous examples. Donald Trump used fake Facebook followers during the US presidential campaign. Researchers discovered that over half of the followers on his page came from countries known as hubs for Facebook ‘like farms’.

These ‘farms’ are often found in developing nations such as the Philippines and India, where the dollar stretches a long way. They offer customers the opportunity to buy fake Facebook likes to create the impression of popular support.

As well as likes, customers can purchase fake engagement, usually in the form of comments. This may explain the unusual commenting activity on Johnson’s page.

For political purposes, this type of artificial campaign is an important tool, because it generates the illusion of grassroots support for a particular figure or issue. It even has a name: astroturfing.

Illusion becomes reality when the fake engagement intersects with genuine users, who are more likely to engage with seemingly popular posts thanks to the effect of ‘social proof’ – a psychological phenomenon where people tend to follow the actions of the masses.

This can be leveraged to great effect in social media environments, where user attention spans are low, knee-jerk reactions are prevalent, and ‘likes’ are an addictive form of currency.

Brexit influence campaign: #BritainIndependence

The atmosphere in Britain is becoming increasingly heated as October 31st inches ever closer. This is the date when the country will leave the European Union — if all goes to plan for Boris Johnson and the Conservative Party. Right now the political stakes are higher than ever.

Parliament has tried to protect the country from the potential catastrophe that may result from leaving without a deal. In the nick of time before prorogation, they passed a last-minute bill into law, designed to compel Johnson to seek an extension, if he can’t reach a suitable deal with the EU by the end of October.

Johnson has already lost multiple times in Parliament, despite his efforts to prorogue it for an unusually long time. Last week, a Scottish court ruled that this prorogation was unlawful. This week, the case goes to the English Supreme Court, where it will be challenged.

#BritainIndependence

In this highly-charged environment, Twitter has provided a constant source of lively political debate around Brexit. Many issues are bubbling up at present, some more relevant than others, but here I want to explore a particularly interesting hashtag.

#BritainIndependence has been popping up a lot lately. The first thing to do is to find out when it was first used, and who first used it. The hashtag came fairly late to the game, on September 9, via a user called ‘Trevor’, whose screen name is stuffed with pro-Brexit hashtags.

Signalling ingroup identity

A quick glance at Trevor’s bio is revealing. First, the bio is a strong indicator of self-professed identity on Twitter. In Trevor’s case, it contains words that reflect traditional values: conservative, nationalist, family-oriented, words such as ‘Christian’, ‘white’, ‘loyal British subject’, and ‘married’.

This creates a sense of group identity, designed to give Trevor’s profile immediate appeal to others who identify with similar values – i.e. signalling that he is part of an ingroup. In this case, the ingroup is pro-Brexit Twitter users.

The references to popular British football teams (Arsenal and Rangers), is likely designed to portray the account as belonging to a genuine working-class British citizen – another effort at signalling ingroup identity.

But the cultural references feel jumbled: Arsenal is an English team, while Rangers is a Scottish team. That plus the random mention of Northern Ireland means this bio doesn’t quite ring true. In fact, it feels like someone playing at being a Brit, perhaps for nefarious reasons.

What’s more, ‘neighbor’ is spelled in US English. No genuine speaker of British English would use US spelling; especially a man who seems so deeply committed to British patriotism. Clue, Trevor is likely not the grassroots British man that he pretends to be.

We could dig much deeper into Trevor’s account, especially his past tweet history. His account is devoted to tweeting about Brexit, even though it was created in September 2015, before Brexit existed. It would be interesting to see what Trevor was tweeting about between then and June 2016, but that’s a topic for another post…

Hashtag activity

Next up, let’s take a look at how Twitter users have been interacting with the #BritainIndependence hashtag, since ‘Trevor’ coined it on September 9th (coincidentally, just in time for the prorogation of Parliament).

All of the most retweeted tweets on the #BritainIndependence hashtag come from users with heavily pro-Brexit screen names (and usernames like @Brexit4me and @Feck_the_EU), suggesting one-topic accounts that exist simply for the purpose of engaging with Brexit-related discussions.

Retweets have two main functions, 1) they spread a message across Twitter, 2) they create validation for the message via social proof (i.e. if other people have engaged with this tweet, then it must be worth engaging with).

Liking (or favouriting) tweets reinforces the sense of social proof, while also increasing the likelihood of the tweet being seen in other users’ feeds.

The below tweets contain strong evidence of typical adversarial narratives, designed to promote a sense of tribalism, i.e. “us vs them”.

Examples include:

  • ‘Unelected judges’ and ‘hijacked by extremists’ (fits into narrative of the EU/Remain supporters being ‘anti-democratic’)
  • ‘Tattooed Eastern European thug’, and ‘brutal rape’ (fits into narrative of foreigners and ‘The Other’ as being threatening, especially to women)
  • ‘Me, just a patriot’ (supports the narrative of pro-Brexit voters as being especially patriotic. This is a similar notion to that conveyed by Trump’s ‘Make America Great Again (#MAGA’), which coincidentally appears in many of the user bios tweeting the #BritainIndependence hashtag.

Clearly, the #BritainIndependence hashtag exists to stoke further divides between the two sides in the Brexit debate, while creating the illusion of widespread support for the pro-Leave side. It’s highly likely that the hashtag was initialised for that very purpose, as the nature of ‘Trevor’s’ account suggests.

Furthermore, it’s telling that this hashtag coincides with several significant real-life events in the Brexit timeline that could threaten the pro-Brexit side, including the beginning of (unlawful) prorogation and the case going to the Supreme Court.

But why are so many of the top posting accounts so similar, with their bios stuffed with tribal keywords ? And why are so many of them blatantly US-centric, or with such obvious cultural errors (such as the spelling of ‘favorite’)?

This could indicate an organised social media manipulation campaign aiming to create the illusion of widespread support for the pro-Brexit side while deepening social and political divisions at a critical juncture for the country.

As October 31 inches closer, the discussion is certain to get even more heated – and we’re sure to see lots of interesting social media activity.

I’ll post further analyses here between now and October 31.

How personal branding paved the way for post-truth

Over a decade ago, an idea was born that seemed innocent at the time, even ground-breaking. It was the idea of personal branding; marketing one’s own skills like a product. In this piece, I’m going to reflect on how the personal branding mindset has played a role in creating today’s polarised and tribal online environment.

In his original Fast Company article,‘The Brand Called You’, author Tom Peters urges his readers to develop their personal brands by delivering talks and developing word-of-mouth marketing around their unique skills. He briefly mentions the importance of showing familiarity with new technology (such as email), but as a rather minor consideration. After all, it was 1997; the digital world hadn’t yet become an inextricable part of everyone’s lives.

Fast forward a few years to the early 2000s, where people had started publishing their own content using blogs and personal websites. The social media platform MySpace was launched in 2003, followed a couple of years later by Facebook. These tools were powerful and they allowed ordinary people to broadcast their message, whatever that might be, to large audiences. It was a whole new way to build the brand called you.

Digital tribalism

The growth of social media and blogs spawned a whole generation of online content creators, some successful, many not. People could now reinvent themselves personally and professionally simply by producing relevant online content and sharing it with audiences via social media. The trick to success was finding a bunch of people with whom your message resonated, i.e. your tribe.

The idea of ‘finding your tribe’ is central in branding strategy, both for commercial marketers and personal branders. Personal branding gurus often stress the importance of being bold and even divisive in the content you choose to publish. The goal in doing so is to eliminate those who aren’t on board with your opinions, leaving only your loyal, like-minded tribe remaining.

Arguably, this tribal approach has instilled in the digital generation a habit of being strongly opinionated online. It’s all too easy to be bold and divisive when you’re safely behind a screen. You can blog, make videos and write ebooks to your hearts’ content.

But creating effective content for personal branding takes up a lot of time and mental energy. Not everyone wants (or has the skills/motivation) to write original blog posts about their key career interests. Luckily, there’s another approach: content curation.

This popular and effortless alternative for building a personal brand community involves sharing other people’s content with your target audience, sometimes (but not necessarily) adding your own quote or original take.

Curation can be done quickly and with the minimum of effort; an appealing strategy in a time-pressed world. For example, content curation on Twitter could be as simple as retweeting articles and tweets relevant to the personal brand you wish to create. By doing this consistently, you can attract like-minded people, which then gives you a tribe, or brand community.

Another relevant factor in the development and solidification of personal branding is the deliberate design of online social networks. This encourages users to take actions which generate more likes, clicks and engagement from their audience.

Content curation and social networks’ design are symbiotic processes intended to complement one another, leading to a cycle where people create (or curate) content, gain approval from their tribe, experience a positive self-esteem boost, and repeat. This ongoing process generates increased traffic for the social networks and more revenue for their vital advertisers.

Personal branding meets politics

In 2008, online social networks made their big debut in politics as part of Barack Obama’s presidential campaign. Obama won the presidency, and followed up in 2012 with another win and another dose of digital political campaigning.

By then, more and more people were using social media and the first signs of manipulation were began to emerge. On top of that, attention spans were beginning to erode as people became used to a lifestyle lived almost wholly online. The introduction of Apple’s iPhone, and the resulting explosion in smartphone use exacerbated this shift, giving people access to social media in their pockets at all times.

It created the beginnings of a world where everyone on a bus or train would have their head down staring at a smartphone. Once we gained the possibility of sharing content at the touch of a ‘share’ button, content curation as part of maintaining a self image would soon become habitual for many. By 2016, social network use was prevalent, most people had a smartphone, and information was flowing non-stop.

Politics had firmly entered the personal branding arena, and campaign managers deployed increasingly clever strategies, such as digital profiling and social ads, to win over voting populations. This practice came to a head with the EU referendum in the UK, closely followed by the 2016 election of Donald Trump as US president.

Going tribal

To better understand what drove these events, it’s useful to first consider the innate human tendency to see the world in terms of ‘us vs them’. This is well demonstrated by the work of behavioural psychologist Henri Tajfel on what he called the ‘social identity theory of group conflict’, in particular the ‘minimal group paradigm’.

The minimal group paradigm shows that people define themselves in opposing groups over the most trivial of matters, such as a coin flip, grouping themselves into ‘Heads’ and ‘ Tails’. Once divided into groups, people tend to favour their own ingroup while disadvantaging, and even derogating, the outgroup. If people can get tribal over a simple coin flip, imagine what they’d be like over political ideology.

Further research has shown that not only do people tend to strongly favour their ingroup, but they also have a tendency to derogate the outgroup. This us vs them mentality manifests in many areas of life, from harmless rivalry over cities and sports (e.g. Boston vs New York or Manchester vs Liverpool), to more serious issues of racism, xenophobia and nationalism.

It also manifests in the digital world, exacerbated by today’s entrenched tendency for personal branding and ‘finding one’s tribe’. People receive positive reinforcement as part of the in-group whenever they broadcast their identity to their fellow brand community members. They usually do this by sharing content, whether their own, or, more commonly, curated from others.

Two infamous political examples are the behaviour of Trump supporters versus Clinton supporters, or Leave versus Remain supporters. Both sides commonly derogate the other (e.g. libtard, Brexiteer) and view their ingroup as superior.

That’s not the only way social identity theory manifests itself in contemporary digital politics. In addition to derogating each others’ perceived political outgroup, it’s become common practice to derogate, and even dehumanise, certain outgroups in wider society, normally minorities such as Muslims, refugees or immigrants.

These groups have become easy targets because of an array of social and political events over recent decades that have put them squarely in the firing line. Ever since the terrorist attacks of 9/11, the British and US mainstream media has consistently highlighted attacks committed by Muslim perpetrators while downplaying similar ones conducted by non-Muslims.

What’s more, the Syrian civil war and the rise of ISIS triggered a massive influx of refugees from Syria and Iraq into Europe. Together, these events produced a climate of fear and uncertainty; fertile territory for innate ‘us and them’ attitudes to thrive in a digital sphere where online tribalism (in the personal branding sense) had long been a common practice.

Tribes before truth

This leads to a very current concern: the rise of online misinformation, often known as ‘fake news’. With such a huge flood of information now available via our smartphones, we don’t always have time to read everything in detail. So we take shortcuts and get lazy about processing information properly. We simply don’t have the time or inclination to think deeply about every piece of content we interact with online.

Nevertheless, we crave engagement and approval from our ‘tribe’. Perhaps we’ve become somewhat addicted to it, to the extent that we sometimes share articles without even reading them. Recent research found that between 50–70% of all URLs on Twitter are shared without being opened, suggesting that people share them based only on the headline. This has heavy implications for the spread of misinformation, and suggests too, that fact-checking probably won’t work.

In an online space rife with misinformation, why would someone share an article without reading it first? Arguably, broadcasting our affiliation to our digital tribe matters more to us than veracity. More critically, broadcasting this affiliation to our ingroup is likely to involve derogation of an outgroup. After all, we really want those likes and shares, and that’s often the best way to get them.

One of the key goals in sharing content on social media (especially Twitter) seems to be to signal that ‘we’ (the ingroup) are different from ‘them’ (the outgroup). This dichotomy shows up most disturbingly in stories about ‘Muslim rape gangs’, refugee ‘sex mobs’, and terrorist attacks that never happened (e.g. the fictitious Bowling Green massacre).

In this tense milieu, it’s easy for misinformation to get picked up and spread as part of the ‘tribal broadcasting’ process, or ‘content curation’ in personal branding parlance. If a certain news story fits people’s ingroup vs outgroup narratives, they’re probably going to share it on social media. Truth may come second to tribalism.

The real danger comes when this digital tribalism plays out in real world scenarios, such as an uptick in anti-immigrant hate crime, or violent events such as ‘Pizzagate’. Both have been linked to online misinformation.

You might ask what the social media giants are doing to address this issue. They’ve made various efforts to implement reporting tools so that users can report hate speech. They have also shut down particularly heinous accounts such as InfoWars, that exist purely to peddle misinformation and hate.

But digital tribalism in fact boosts all the metrics that spell success for social media firms, creating a self-reinforcing situation. One can’t help but wonder how far they’ll actually go to rein this in.

If only we could all quit social media, en masse. Would that solve the problem? Or does it run deeper than that?

Tweeting my way into academia

On Twitter not long ago, someone suggested that academics should avoid using social media. He cited reasons such as distractions, narcissism, and ‘the risk of getting trolled’.

I’m studying social media for my PhD and so I’m well aware of its flaws. But, like any tool, when used in the right way it can open up many new opportunities. Here, I’ll speak out in defence of Twitter’s usefulness, and explain how I used it to find an unadvertised, fully-funded PhD opportunity.

My professional life has revolved around Twitter for over six years, ever since 2012 when I started blogging about nation branding. It helped me gain traction networking in the field, which led to media interview requests, conference and keynote speaking invitation in far-flung destinations (Jamaica, Jordan, Turkey, Indonesia…) and assorted consulting gigs.

That blog also played a central role in my eventually securing a competitive job at a London software startup. All because of a blog and Twitter.

It wasn’t that complicated. I simply wrote posts, stuck them up on my website and then publicised them on social media along with the requisite hashtags. I interacted with people who replied, and took the conversation in interesting directions.

But the contacts I made were invaluable, and, perhaps more importantly, Twitter was central in enabling me to join conversations around the topic of nation branding, get my thoughts out into the world, and in the process build up my expertise and credibility.

This PhD may be the biggest victory to date for my Twitter use. It all started back in early 2017, when I was working in London. It was a good experience at an exceptional company. I learned a lot about how tech startups work, and I enjoyed the time spent with my colleagues.

But I just wasn’t passionate enough about the subject matter and I yearned for something more. For a long time, ever since my time in Istanbul, I’d been deeply intrigued by politics, international relations and media, often with dashes of technology around the edges.

At around the same time, the unfolding saga of Brexit, Trump and online radicalisation captured my attention in a big way. Social media played a key role in the story. There was much talk about ISIS using social media platforms to brainwash vulnerable young people and entice them to Syria.

As 2017 slipped by, the primary narrative around social media and politics shifted. It began to focus less on radicalisation and more on how various foreign influences (and perhaps homegrown ones too…) had used social media to foment dissent against the status quo. Some even argued that our democracy itself was being subverted, hijacked by bad actors.

It was propaganda for the digital age and it fascinated me. I wanted to study it in more depth. Academia offered the perfect platform to do that.

Having no desire to take out loans or decimate my savings, I knew I had to find a PhD that offered full funding. I applied to an advertised position at the University of Sheffield, to research the role of visual social media in fake news.

After reaching the interview stage, they told me I hadn’t been selected for the role. Nevertheless it was a useful experience, because now I had a whole PhD proposal ready to go. It just needed to find a home.

One day in the aftermath of that minor setback, I was browsing Twitter, looking through hashtags relevant to my interests, seeing if any opportunities might pop up. Those hashtags led me to a professor at the University of Bath, who was researching cybersecurity and online trust.

I pinged him a DM, explaining that I had a proposal that could be relevant to his research interests, and asked if he’d be keen to take a look. He was, so we Skyped and exchanged emails in which he advised me on how best to fine-tune the proposal.

I got up at 5am every day that week to get it ready for submission. The university accepted it and miraculously there was full funding available.

I quit my startup job, and the rest is history. I’m now happily immersed in a research topic that I find meaningful, while also developing new skills in Python, network analysis, machine learning and statistics.

I’m eventually planning to go back into industry rather than continue in academia, but these skills will be invaluable whichever route I choose to take.