Nature paper says bad news is good news

‘Negativity drives online news consumption’, Claire E. Robertson et al., Nature Human Behaviour, March 16, 2023:

Here we analyse the effect of negative words on news consumption using a massive online dataset of viral news stories from—a website that was one of the most successful pioneers of click-bait in the history of the Internet23.

The tendency for individuals to attend to negative news reflects something foundational about human cognition—that humans preferentially attend to negative stimuli across many domains24,25. Attentional biases towards negative stimuli begin in infancy26 and persist into adulthood as a fast and automatic response27. Furthermore, negative information may be more ‘sticky’ in our brains; people weigh negative information more heavily than positive information, when learning about themselves, learning about others and making decisions28,29,30. This may be due to negative information automatically activating threat responses—knowing about possible negative outcomes allows for planning and avoidance of potentially harmful or painful experiences31,32,33.

Previous work has explored the role of negativity in driving online behaviour. In particular, negative language in online content has been linked to user engagement, that is, sharing activities22,34,35,36,37,38,39. As such, negativity embedded in online content explains the speed and virality of online diffusion dynamics (for example, response time, branching of online cascades)7,34,35,37,39,40,41. Further, online stories from social media perceived as negative garner more reactions (for example, likes, Facebook reactions)42,43. Negativity in news increases physiological activations44, and negative news is more likely to be remembered by users45,46,47. Some previous works have also investigated negativity effects for specific topics such as political communication and economics34,48,49,50,51,52.

The framing here is curious. The paper’s authors hypothesised that readers prefer bad news (determined by words signalling negativity in the headline) based on the fact that bad news is “more ‘sticky’ in our brains”, because people use bad news to know how to plan ahead. They further justify their choice – a reasonable one, to be sure – based on the effects of negativity on readers’ news recall and physiology.

Second, after their study seemed to proved their hypothesis, the authors delved into which negative emotions invoked stronger interest from readers:

… high-arousal negative emotions such as anger or fear have been found to efficiently attract attention and be quickly recognizable in facial expressions and body language31,59,60. This may be because of the social and informational value that high-arousal emotions such as anger and fear hold—both could alert others in one’s group to threats, and paying preferential attention and recognition to these emotions could help the group survive27,32. This may also be why in the current age, people are more likely to share and engage with online content that is embedding anger, fear or sadness21,41,61,62. Therefore, we examine the effects of words related to anger and fear (as high-arousal negative emotions), as well as sadness (as a low-arousal negative emotion).

Across both analyses as well as throughout the paper, there seem to be two implicit assumptions: a) that news producers have an option to choose between good and bad news (loosely defined), and b) that the network of outlets producing mis- and dis-information in the guise of ‘news’ has no role to play.

On the former: Journalists often don’t. In fact, good journalism in a country with a government that fixates on some arbitrary ‘good news’ even at the worst of times needs to counter-fixate on what national leaders are keen to ignore. And the news reports produced thus are not this way to encourage click-throughs but because they are, in fact, what’s happening.

Of course, the authors’ data source is limited to the content published by and its readers in the US, which limits what conclusions they can reach – but it also leaves ambiguous whether their not reaching certain conclusions is due to these limitations, because it didn’t fit their hypothesis or because they didn’t even consider the possibility.

On the latter: The authors only admit that it’s important to understand “the biases that influence people’s consumption of online content … especially as misinformation, fake news and conspiracy theories proliferate online” – but even here suggesting that journalists wield negativity as a way to manipulate people into consuming the news. Throwaway lines like “Even publishers marketed as ‘good news websites’ are benefiting from negativity” aren’t helping; what is this supposed to mean anyway?

Similarly, the ‘Discussion’ section of their paper concludes thus:

Knowing what features of news make articles interesting to people is a necessary first step for this purpose and will enable us to increase online literacy and to develop transparent online news practices.

This sentence spells out a logically straightforward premise that suggests we can use the study’s results to look through or past headlines that seem to say something but actually say nothing at all (you know these). Yet it is also infused with an assumption that readers’ news-consumption habits don’t influence journalists’, and newsroom-runners’, choices.

I’m not saying “we’re just responding to readers’ demands”; instead, I’m saying that news consumption is a combination of publishing and consuming habits. Like news publishers are competing for readers’ attention, readers are often competing to discover the ‘best’ or a ‘better’ news source, especially on topics also associated with considerable misinformation (e.g. vaccines), and which – as studies based on Facebook use have demonstrated – they can admit into their echo chambers. We need clarity on how they make their choices in this process of discovery, and why, in different parts of the world.

Without this knowledge – which the authors appear to have sidelined based on the idea that journalists have an option to not be negative – we’ll never “develop transparent online news practices”.