The Twitterbot Effect

28/12/2016 11:44 GMT | Updated 28/12/2016 14:48 GMT

In 2009, Jack Dorsey, CEO of Twitter Inc. said this; "...we came across the word 'twitter', and it was just perfect. The definition was 'a short burst of inconsequential information,' and 'chirps from birds'. Twitter is simple, potentially informative and dedicated to free speech before anything else. Today, there are approximately 316 million people with accounts. Today, the PEOTUS uses Twitter as a personal confessional to his nation-in-waiting. Radicalised Islamist group ISIS is estimated to have 46,000 accounts on the site. A moth caught on camera landing on Cristiano Ronaldo's face at Euro 2016 has over 50 accounts to its name. Twitter's unique brand of microblogging gives us the accessible, simplified and instantaneous newsfeed we wanted before, and rampantly use now. Trends on Twitter rarely tend to evade our attention. Most of it is nonsensical, some pervasively important. We are advertised, appealed to and informed by the millions of users who actively "tweet" approximately 6,000 times a second.

Within this quivering mess of opinion, debate and argument, based on most estimates, around 23 million of those users aren't human. Generated by computers and paid for by people, known as "twitterbots"; even Twitter cannot be sure of how many accounts like these exist. They can generate apparent mass support for causes under the noses of genuine users.

What are Twitterbots?

Existing in a peculiar pergatory between spam and free speech, Twitterbots vary significantly in their sophistication. Some are basic, posting automated responses triggered by certain keywords or phrased of actual tweets. For example, "@mothgenerator" responds to an "@" tweet by translating the text sent to it into numerical data and creating, through a complex JavaScript process, a unique, imagined moth complete with concept image and species name. This example might not strike fear into the heart of many, unless you suffer from a fear of moths. Not all bots are as harmless. The New York Times explained how some Twitterbots simulate "sleep-wake cycles" to make them appear more convincing, and harder for software to detect and remove them. These bots can post anything from targeted product advertising to flagrant slander and misinformation on a massive scale. They can target certain accounts to smear certain people, or focus posts on those who companies believe will be influenced by certain advertising based on their previous interests.

It has been possible for several years, on most social network platforms, to purchase accounts with a specific intention of increasing follower counts or support for certain opinions. This doesn't have massive impact in the context of isolated cases. When Twitterbots begin to influence elections, brands and shares, the real threat becomes clear.

Why this matters

According to Twitter, approximately 24% of tweets originate from bot accounts. The Science Direct journal conducted an investigation; finding that humans were surprisingly receptive to bot tweets, perceiving them as both credible and competent. Essentially, Twitterbots, when carefully constructed and deployed can operate unnoticed and effectively as a human agent for anything from advertising to trend-setting.

Twitter had a valuation of around $8.4 billion dollars as of 2011; it will have soared since then. Its social power and impact is staggering in terms of the level of attention it currently holds sway over. Therefore, even as little as 5% of its account users being automated computer generators designed specifically to shape opinion, ravage reputation or make money should be concerning.

The US presidential election was won, in some part, thanks to Donald Trump's actions on social media platforms. Bloomberg reported on a study conducted into specifically Pro-Trump accounts and their tweeting habits over an eight-day period. They found that of the 19.4 million tweets, approximately 18% were from bot-related accounts (tweeting 50 times a day with the same hashtags). The most active bot accounts were operating at an output 2,600 higher than a normal one, with 1,300+ tweets over 24 hours. Douglas Guilbeault, who conducted the study claimed that "If bots have even a minor influence, that's scary".

In a world where fake news can trigger international events, the notion that Twitter suffers from manipulation like this should scare us. Knowledge of "Twitterbots" peaked a few years ago, when most articles were published on the topic. Twitter is privately owned, money-making entity; it wouldn't entirely unbelievable to think that they would be reluctant to totally eradicate bot activity, if it meant damaging the size of their apparent community, which defines their financial value. The impact bots can have is hard to accurately ascertain, but if we genuinely believe in transparency, integrity and trust as pillars of western society, it becomes all the more pertinent to focus on routing manipulation like this out of our most vital platforms of free speech.