We are starting to have some suggestion about what kind of content is advantaged on social media, all other things being equal. Good candidates are negative content, generally emotional content, or out-group detraction (link to some papers if you click on the contents).
These effects, albeit generally quite small, are interesting, and they are consistent with the broadly accepted narrative that social media have a detrimental influence on human communication and social interactions. This may definitely be true, but what I believe is an important missing piece is a contrast with other forms of communication and social interactions.
In other words, how do we know that it is because of social media that negative content - or emotional, or partisan, etc. - is favoured? To answer this question, we would need to isolate the features of social media we suppose are behind the effect, and compare situations in which these features are present and not, and we should be able to show that, say, negative content is favoured when the features are present, and not when they are not.
We can consider other forms of communication, and see what are the features that differ with respect to social media-mediated communication. An obvious starting point is oral, face-to-face, communication, a form of social interaction that has been dominant for the majority of human history (and partly still is). There are various studies in cultural evolution, for example, that show how emotional or negative content is favoured also in experiments that reproduce situations similar to oral transmission. How does oral communication differ from social media-mediated communication? First, anonymity is, though not impossible, difficult to implement in oral communication, while it is easier to implement in social media. Second, oral communication generally takes place in small group, where people often know each other. Third, the oral back-and-forth requires understanding, memorisation, and (re)production, while in social media one just need to decide what to share (understanding itself is not required if memorisation and reproduction are not needed).
Now, there could be other interesting differences (suggestions welcome!), but let’s stop here. I recently published a paper where I tried to do something similar, focusing on the last difference. I used “attractive” content that was found to be advantaged in recall and transmission in cultural evolution experiments - negative, eliciting disgust, and threat-related - and I compared it to a “neutral” version of the same content, i.e. without the elements making it negative, etc. The first, experiment, let’s call it “storytelling” scenario, was a situation resembling oral transmission (participants had to listen - actually read on a computer screen - a story and then retell - write - it to another participant, and so on) and the second one, “social media” scenario, was resembling online sharing (participants had to read a story and decide whether they would share it or not).
Long story short, the results presented a mixed picture (experiments and analyses were fully preregistered, maybe there is a connection here?!), but two suggestions can be summarised here:
Negative content was the only content that was both better recalled and transmitted in the “storytelling” experiment and shared in the “social media” experiment with respect to its neutral counterpart.
Overall, attractive contents seem to be more advantaged in the “storytelling” than in the “social media” scenario. On one side, this seem reasonable, if we think that this content is attractive because of its properties, and its properties impact on understanding, memorisation, etc. On the other side, is feels surprising as we tend to believe that social media communication is more biased than oral one. (While I like this result, it needs to be said that the kind of content was chosen because advantaged in experiments resembling oral transmission.)
All data and code to reproduce the analysis are available here.
To conclude: as usual, things are complicated, but I do think that comparing social media communication with other forms of communication is the only way to understand the possible causal effect of social media. In addition, if we could isolate the undesirable features (if any), we could design specific, targeted, interventions.