Emotion trumps reason?

This is how the 45th and 47th President of the United States ended his tweet sent at 1.42am on the 19th of December, 2020. The tweet is now considered to have been instrumental in not only bringing protesters to Washington D.C. on January 6th, 2021, but in fomenting the violence which occurred that day. In the words of Jamie Raskin, a member of the House select committee investigating the Capitol riots:

"Donald Trump would issue a tweet that would galvanize his followers, unleash a political firestorm, and change the course of our history as a country".

That a single post on a website could be considered so consequential for democracy speaks volumes about how social media have revolutionised the way we talk about politics. Trump is widely considered to be an expert X (FKA Twitter) communicator, and regularly uses emotive rhetoric in his tweets. Indeed, one of his final two tweets before being banned in January 2021 would read:

Is he an outlier, or is this a 'new normal'? In other words, have politicians become more emotional in the language they use on social media? Prior to Musk's takeover (and, crucially, before he put Twitter's API behind an enormous paywall) I examined 8 million tweets posted by British MPs from 2007 until the General Election in 2019. Yes, even the Ed Balls tweet is in there. In a nutshell, I found a small but steady increase in the levels of emotion used by politicians over time. Trump might be the absolute apex of this trend, but it does seem like a trend.

Why might this be happening? First, I had to discount things like age, gender, or the emotions attached to different topics over time. 2010-2020 was a famously mundane decade in British politics, so I had to make sure I wasn't just picking up the effect of Brexit or Scottish independence. I could control for a range of variables and, using topic modelling, even the main issue of each tweet. The effect persisted. What I did find was a statistically-significant relationship between negativity and and the amount of engagement (likes, retweets, quote tweets and replies) their tweets received. Pre-Musk, Twitter's algorithm was pretty transparent, with more engagement generally resulting in greater visibility. I suggest that politicians learned to employ increasingly emotional rhetoric to maximise the reach of their messages and raise their electoral profiles.

What might we take from this? First of all, even in the (relatively) benign pre-Musk age, the design choices made by social media sites have profound consequences for our behaviour - and the messages we see. With democracy seemingly in retreat around the world, this promotion of emotive content has real consequences for the health of political deliberation. Emotive tweets are intended to influence behaviour and help elites achieve their political goals. Affective language taps into primal instincts, reinforcing pre-existing beliefs and deepening both ideological and emotional divides. We also know that evoking emotion can motivate people to engage in a wide range of political activities, whether that is simply getting people to the polls, or inciting them to engage in protest or violence.

Normatively we tend to think that healthy democracies should strive for more reason and less emotion in the pursuit of better policy-making. Reason promotes objectivity and impartiality in fair and just decision-making, allowing allows policy-makers to evaluate evidence, consider diverse perspectives, and use logic and facts rather than personal biases or transient feelings. Heightened emotions also exacerbate problems in polarised societies. Emotionally-charged traditional media have historically framed issues in ways which evoke strong, polarised reactions, simplifying complex topics into binary `us-versus-them' narratives. It seems as if we are witnessing the emergence of similarly sensationalist and divisive political discourse online.