How to improve social media - Part I
The benefits of looking for consensus
The intended third and final post in my series on how to improve debates in the public sphere became long enough that it warranted two posts. This is the first one on how we could, in principle, improve social media platforms. The next post will look at how these insights could be implemented in practice.
“For many years there was a feeling that the wonderful things on the web were going to dominate and we'd have a world with less conflict, more understanding, more and better science, and good democracy". This view, expressed by Tim Berners-Lee, the inventor of the World Wide Web, has certainly been shared by many. The Internet had the potential to make the public sphere more democratic and to improve the exchange of ideas. This perspective was echoed by Elon Musk, the new owner of Twitter, when he described the platform as “the digital town square, where matters vital to the future of humanity are debated”.
Unfortunately, while the Internet has created a global public sphere, the result has been somewhat mixed relative to expectations. The same Tim Berners-Lee conceded that “people have become disillusioned because of all the things they see in the headlines”. The Internet, and in particular social media, is thought to have generated online fights and polarised crowds. In this post, I look at how we could improve social media for them to be better spaces for exchanging ideas so that the best ones prevail.
The example of Twitter and its flaws as a digital town hall
As a digital town hall, Twitter seems to generate less-than-ideal types of interactions. Reactions such as anger, outrage, mockery, and pile-ons appear much more frequently than in face-to-face interactions.
Many users, while staying on the platform, describe its atmosphere in negative terms.
One example of a Twitter feature that contributes to the negative atmosphere is the “quote-tweet,” where users can retweet a post with their comments on top. These are frequently used to point to tweets one finds ridiculous or morally appalling, and to use them as exhibits of the stupidity of others, adding ironic or outraged comments.
Overall, the outcome of the global “discussion” on Twitter is questionable. It has raised a lot of concerns about polarisation, with users’ discussions being feared to be clustered around echo chambers of like-minded people with opposite views. Research on this question has shown that the problem of echo chambers and polarisation on social media is likely not as dramatic as often portrayed (Guess, 2018; Ross Arguedas, 2022). People seem to access potentially more diversified political content on social media than outside, and the proportion of people isolated in echo chambers is somewhat small (typically less than 10%).
That being said, it may be easier for small communities of like-minded people with extreme views to form now on social media than previously offline. Members of such communities may have a disproportionate impact on public discourse with their higher investment in contributing to public discussions.
Status games and their incentives
Noticeably, not all social media platforms are characterised by negative interactions. Indeed, different social media platforms seem to have very different cultures of interaction. This difference in culture across platforms is pointed out in memes suggesting that the same people adopt very different attitudes and social personae on different platforms.
Games of status
To understand why the same people would behave so differently, we have to consider why and how social platforms incentivise users to engage and post on them. Clearly, it is not money: most users do not get financial rewards from posting.
Social media are games of status, that is, social settings where people can compete to climb a ladder of social recognition.1 The currency with which social media users are rewarded is the likes and reposts of their contributions, which signal the social approval of others. Their social standing in a social media game of status is the social score associated with their profile, such as their number of followers on Twitter, Facebook, or Instagram, and their Karma score on Reddit. Higher social standings on social media platforms provide prestige, a specific type of status. “Prestige is rooted not in coercion but in admiration, respect, and voluntary deference.” (Dan Williams)
Receiving attention and approval on social media is clearly a driver of engagement. A study found that people space their engagement on social media to maximise their social rewards, such as likes and retweets (Lindström et al., 2021). Another study found that after going viral, posters “more than double their rate of content production for a month” (Srinivasan, 2023).
Incentives
The reason for the different attitudes on different social platforms is that different games of status can be structured around different rules of play that emerge in social communities as social norms. These norms determine what gets rewarded and what gets frowned upon. For instance, the professional network LinkedIn features overwhelmingly positive posts about professional achievements and encouraging wisdom. Users there often acknowledge that negativity is a kind of faux pas. On the contrary, on Twitter, expressions of outrage have been found to increase positive social feedback, thereby increasing a user’s propensity for future outrage expressions (Brady et al., 2021).
So people react to incentives provided by the different social norms of social media platforms to produce content and engage with others in the search for social recognition. The understanding that people have such incentives paves the way to a new question: can we shape these incentives so that people engage in behaviours that are socially beneficial instead of socially detrimental?
One of the most interesting projects in institutional design is to investigate the conditions under which the intangible hand [of the economy of esteem] can be expected to work well and, in particular, to work for the production of what is generally seen as the common good. - Brennan and Pettit (2005)
The social norms on different social media platforms are not random; they emerge as one of the possible solutions (see this post) of how these games could be played given their features. Some of the features of social media that can influence prevailing behaviour include which posts are amplified by algorithms and how posts are rewarded/supported by other users. To change the norms of social media platforms—with the aim of creating a better public sphere—we can think of the key features of social media that could induce less partisan and conflictual behaviour and more constructive and reasoned contributions.
Favouring consensual views
How to get over the partial and partisan perspectives
The ideal vision of a public sphere, where public opinion is formed, portrays a marketplace of ideas where the best concepts prevail. A challenge to this vision arises from the fact that people are not typically motivated to seek the truth, engage in thoughtful discourse, or present the most rigorous arguments. Instead, they often strive to persuade others that their standpoint, usually aligned with their self-interest, is correct, and they seek the most persuasive rhetorical strategies to achieve this goal. Consequently, individuals tend to search for conveniently supportive narratives and ideas. Rather than functioning as a marketplace of ideas, our public sphere often resembles a marketplace of rationalisations.
From this perspective, improving debates on social media seems a daunting challenge. How can we ensure that evidence-based and logically rigorous arguments prevail when the majority of individuals primarily seek convenient viewpoints? One potential solution involves appointing arbitrators to assist in determining the validity of arguments. However, arbitrators, being humans themselves, are susceptible to personal incentives and biases, raising concerns about their ability to perform their role objectively.2
In the absence of objective arbitrators, how can the public sphere effectively select the best ideas from participants motivated to seek rationalisations? This task appears as challenging as the fantastical endeavour of the Baron of Munchausen, who famously extricated himself from a swamp without needing to step on anything solid, simply by pulling on his own hair.
The kernel of truth in individual perspectives
A solution to this problem can be found in the recognition that different perspectives often harbour a kernel of truth. This notion is illustrated in the famous fable of the blind men and the elephant, wherein blind individuals develop very different conceptions of the same elephant based on their exploration of different parts of its body.
Hence when people with different perspectives happen to agree on something, we can think that it is where these perspectives are most likely to be compatible with the truth out there.
The notion that a view is more likely to be correct if a large number of people share it is supported by a mathematical result, Condorcet's Jury Theorem. If each member of a group of people possesses slightly better than a 50% chance of holding a correct view on a particular issue, then the probability of the majority of the group arriving at the correct conclusion is much larger than 50%.3
Another noteworthy observation is that the aggregation of multiple perspectives frequently serves as an effective method for approximating the truth. This phenomenon was famously demonstrated by the statistician Galton. At the 1906 West of England Fat Stock and Poultry Exhibition, attendees were encouraged to estimate the weight of an ox before its slaughter. Galton gathered estimates from 800 participants. Despite considerable individual variations in the estimates, he discovered that the average estimate was 1,197 pounds, a value remarkably close to the actual weight of 1,198 pounds.
In other words, the crowd's judgment was essentially perfect. […] Galton wrote later: "The result seems more creditable to the trustworthiness of a democratic judgment than might have been expected." That was, to say the least, an understatement. - Surowiecki (2004)
James Surowiecki’s bestseller The Wisdom of Crowds detailed the numerous ways in which the aggregation of individual perspectives can enhance our ability to make informed and improved decisions. The wisdom of crowds points to us a way to identify better ideas: those that have a broader appeal.
Partisan perspectives and truth
The preceding argument for the wisdom of crowds may seem be overly optimistic. It somewhat assumes that people's varying perspectives arise primarily from random factors rather than differing interests. In examples such as the fable of the elephant or Galton's guessing game, the individuals involved ideally share a common objective of seeking the truth. However, in the context of political debates, contrasting viewpoints often originate from varying interests and motivated reasoning, rather than mere randomness.
If people’s perspectives are shaped by their interests, one may question whether there should be any kernel of truth in different perspectives. Why would the centrifugal force of interests not fully disconnect individual perspectives from objective reality? Why would people simply not harbour the most convenient delusions?
It is a fair question. The answer to this is that people do not form these views for their own enjoyment, but to convince others (Mercier and Sperber, 2011). To be convincing, these views cannot be entirely nonsensical. The arguments more clearly grounded in unambiguous evidence and more straightforwardly logically consistent are more likely to hold up to scrutiny and stand undefeated in debates. For that reason, debates tend to pit arguments that feature some aspects of a common reality.
Consider a wealthy landowner arguing that land taxes should be decreased. He will be more convincing if he has a theory about the adverse effects of taxation supported by factual evidence and solid reasoning. What limits the centrifugal force of the landowner's interests away from reality is the fact that his arguments have to withstand the scrutiny of others.
Therefore, even when people’s perspectives are heavily shaped by their divergent interests, the few ideas on which these people agree are more likely to be consistent with evidence and logic. We can leverage this insight to use broad appeal across partisan perspectives as a criterion signalling greater credibility. Unlike the simple wisdom of crowds principle, better ideas are not necessarily those endorsed by a large number of people (who could have the same partisan perspective); they are those that garner support from people with diverse interests and partisan views.
We then have our criterion to extract ourselves from the swamp of nonsensical and self-serving ideas: consensus across partisan lines. This criterion can be used by social media algorithms to improve the quality of their content. Even though people follow their interests on social media, we can aim to shape the incentives of these platforms to reward content that is more consensual and receives support across partisan lines.
Can we get good ideas out of a crowd of people motivated to push self-serving narratives? I believe so. We can use this same crowd as judges of ideas and the criterion of consensus as a measure of the quality of ideas. From this idea stems a possibility to improve public debates on social media platforms: designing their incentives so that contributors are rewarded for gaining consensual support for their views, across partisan lines.
In her famous book Governing the Commons (1990), the economist and Nobel Prize winner Elinor Ostrom explained how we can design institutions so that cooperation emerges in social settings where people are mostly strategic and self-interested. Looking to design social media’s incentives, so that the best ideas prevail among people who may not be seeking the truth, can be seen as an extension of her approach (Farrell et al. 2023).
In the second part of this post, I will look at the practical solutions (e.g., algorithms, rewards) to incentivise social media users to produce content with broad appeal across partisan lines, and at the existing empirical evidence in favour of these solutions.
References
Brady, W.J., McLoughlin, K., Doan, T.N. and Crockett, M.J., 2021. How social learning amplifies moral outrage expression in online social networks. Science Advances, 7(33), p.eabe5641.
Farrell, H., Mercier, H. and Schwartzberg, M., 2023. Analytical Democratic Theory: A Microfoundational Approach. American Political Science Review, 117(2), pp.767-772.
Guess, A., Nyhan, B., Lyons, B. and Reifler, J., 2018. Avoiding the echo chamber about echo chambers. Knight Foundation, 2(1), pp.1-25.
Lindström, B., Bellander, M., Schultner, D.T., Chang, A., Tobler, P.N. and Amodio, D.M., 2021. A computational reward learning account of social media engagement. Nature communications, 12(1), p.1311.
Mercier, H. and Sperber, D., 2011. Why do humans reason? Arguments for an argumentative theory. Behavioral and Brain Sciences, 34(2), pp.57-74.
Ostrom, E., 1990. Governing the commons: The evolution of institutions for collective action. Cambridge University Press.
Ross Arguedas, A., Robertson, C., Fletcher, R. and Nielsen, R., 2022. Echo chambers, filter bubbles, and polarisation: A literature review.
Schwartzberg, M., 2015. Epistemic democracy and its challenges. Annual review of political science, 18, pp.187-203.
Shubik, M., 1971. Games of status. Behavioral Science, 16(2), pp.117-129.
Srinivasan, K., 2023. Paying Attention. Technical Report 2023. mimeo, University of Chicago.
Surowiecki, J., 2004. The Wisdom of Crowds/James Surowiecki. NY.: Anchor.
I have discussed this point in the case of fact-checkers in my last post.
The fact that everybody has more than a 50% chance to be correct can be seen as the reflection that their views are influenced by a common reality, which provides them with somewhat informative signals about what’s likely to be true.
Nice approach! I hope you touch on the incentive structure for the social media companies themselves. Tribalism seems to drive engagement, aiming for another equilibrium would require some kind of incentive structure that encourages platforms to make a pivot. (Or at least a story for why a platform would chose to move towards a new equilibrium, given market pressure and the multi-polar situation they're in)