That's a fascinating study, with sobering conclusions. I hadn't thought of it as a matter of wanting to stand out. Perhaps that's part of the enduring appeal and continuous churn of counterculture...
That's a fascinating study, with sobering conclusions. I hadn't thought of it as a matter of wanting to stand out. Perhaps that's part of the enduring appeal and continuous churn of counterculture groups? What it says to me is that there's benefit in having spaces outside of the mainstream where people can go and express their differences purely for the feeling of being different, even aside from any actual differences.
My solution: We need more (benign) cults, secret societies, hidden libraries, and other forms of non-toxic differentiators and social groups that don't harm the people involved, which can inoculate them from the appeal of anti-establishment groups that are more harmful to society and to the people involved.
Your suggestions all sounds like a lot of fun actually, just imagining them makes me feel like wanting to find and join them, I think you're on to something here Anecdotally, my experience is...
Your suggestions all sounds like a lot of fun actually, just imagining them makes me feel like wanting to find and join them, I think you're on to something here
Anecdotally, my experience is communities that focus on working toward a goal or completing a project also tend to be more supportive. Maybe something about being part of a higher cause softens the desire to stand out
Fan fiction and conspiracy as well. The appeal of something made up that we can invest in and be unique easily. Expend our extra mental cycles. Idle hands. Even when I was forming this post I...
Fan fiction and conspiracy as well. The appeal of something made up that we can invest in and be unique easily. Expend our extra mental cycles. Idle hands.
Even when I was forming this post I searched for 'conspiracy' in the comments to see if it was novel. In this case I would have self-censored had someone mentioned something different. But my first thought was self-censor, not negativity.
Man this study is awesome and it brushes up against some of my own personal hyperfixations such as online propaganda and online tribalism. I think there's an even deeper psychological connection...
Man this study is awesome and it brushes up against some of my own personal hyperfixations such as online propaganda and online tribalism.
I think there's an even deeper psychological connection here between differentiation, identity, perceived superiority, and hostility.
There's a concept in social psychology called "positive distinctiveness," which describes how people don't necessarily just want to be different from others, they want those differences to reflect positively on themselves and/or their in-groups. People generally want to view their identities, beliefs, and social groups favorably, and one way of accomplishing that is through another concept called "social comparison."
And one of the easiest ways to establish a sense of superiority isn't necessarily to demonstrate your own intelligence, morality, or competence, but to portray someone else as less intelligent, less moral, or less competent.
That's where I think this study gets particularly interesting.
People increasingly tie their opinions to their identities. An opinion stops being something they believe and becomes part of who they are. Consequently, someone challenging that opinion isn't necessarily perceived as presenting an alternative perspective, but as threatening their identity and their positive perception of themselves.
This connects to the social psychology concepts of "identity-protective cognition" and "motivated reasoning," where people become more inclined to defend information that reinforces their identity and reject information that threatens it, regardless of its factual accuracy.
Now combine that with the desire for differentiation demonstrated in this study.
People want to stand out, but they also want to feel good about what makes them different. Negativity offers an accessible way to accomplish both. Criticizing someone else's opinion can simultaneously distinguish your contribution, establish perceived intellectual or moral superiority, and reinforce your identity within a social group.
And social media can reward this behavior through attention, engagement, and validation from like-minded people.
There's also a fascinating connection to Optimal Distinctiveness Theory, which proposes that humans have competing needs for individuality and belonging. Online tribalism can satisfy both simultaneously. You establish your individuality by identifying with a group whose beliefs distinguish you from outsiders, while gaining belonging and validation from people who share those beliefs. While conflict with others in the outside groups reinforces those beliefs.
My hyperfixation on online propaganda is actually what led me to explore these psychological mechanisms in the first place, because propaganda exploits precisely the motivations and vulnerabilities we're discussing.
I have found a particularly effective propaganda strategy isn't simply convincing someone that a particular claim is true. It's getting them to incorporate that claim into their identity, especially by associating it with being smarter, more informed, more moral, or more enlightened than people who disagree.
What this means is that a lot of disinformation and propaganda isn't just framed as "information", it is ALWAYS spun with emotionally charged language and frames it in the context of "in-groups" and "out-groups", even if those groups are just ideologies and not specific groups.
Once that happens, contradictory evidence can become psychologically threatening. Accepting that you might be wrong no longer means merely reconsidering an opinion. It can mean questioning something you've used to define your identity, your social belonging, and potentially your sense of superiority.
And when attacking opposing viewpoints earns social validation, the hostility itself can become reinforcing.
What's especially interesting is that these mechanisms don't necessarily require people to be consciously seeking superiority, and they can operate across ideological and political boundaries.
Obviously, this study doesn't establish that entire process. It specifically demonstrates a connection between the motivation to differentiate and increasing negativity. But when you combine its findings with existing research on social identity, positive distinctiveness, social comparison, motivated reasoning, and status-seeking behavior, I think there's a compelling larger hypothesis worth investigating.
Social media may not simply amplify existing tribalism and hostility. Its incentives for differentiation and social validation may actively encourage people to build identities around increasingly oppositional beliefs, then reward them for defending those identities through antagonism.
And that creates an environment in which propaganda can potentially become especially effective, because people aren't just defending what they believe. They're defending who they believe themselves to be.
Woah I haven't thought of how all the pieces fits together like this, thanks. I know personally it's quite difficult to take someone critquing an argument I agree with not as also an attack on my...
Woah I haven't thought of how all the pieces fits together like this, thanks. I know personally it's quite difficult to take someone critquing an argument I agree with not as also an attack on my belief, the dark pattern is hard to avoid.
Link to preprint (free to read version). Excerpts: Introduction: [...] Discussion (findings): Edit: So I suppose the question now is how can we differentiate ourselves without resorting to negativity.
Link to preprint (free to read version). Excerpts:
Introduction:
One of the earliest theories of online discourse was Godwin’s law–the adage that, as an online discussion grows longer, the probability of a comparison involving Nazis or Hitler approaches one. While Godwin’s law does not appear to be literally accurate (1), it also raises a broader prediction: that online conversations tend to devolve into negativity over time. Here, we examine this prediction by analyzing whether social media discourse becomes more negative as discussions evolve, and testing why this happens.
[...] Most social media users believe that online platforms are divisive, toxic, and have a harmful impact on society (7,8). One leading explanation of this negativity is that social media incentivizes negative content because it elicits more attention and engagement than positive content (5,7,9). Engagement-based algorithms therefore amplify negative posts, pushing them to the top of users’ feeds (3,10,11). This visibility may lead users to internalize norms of negativity or strategically post negative content to attract attention themselves. In this model, users who post negative content seek to do so–either to conform to platform norms or to attract engagement.
Here we explore why users may also post negative content as a byproduct of a more benevolent motivation: the desire to differentiate oneself. Psychologists have long recognized differentiation as a fundamental human motive (12), which influences group behavior in offline contexts (13). However, social media accentuates this motivation by rewarding unique posts with shares, likes, and views (7). [...] Ironically, however, the structure of social media also makes it hard for users to stand out. Social media threads involve sustained discussion of a single topic. As time passes and more people comment, users may find it increasingly challenging to make unique contributions to these discussions and draw attention to their comments.
We propose that, as users find it increasingly difficult to stand out, they may resort to negativity as a means of differentiating themselves. Negativity facilitates differentiation for two reasons. First, negative information is more semantically heterogeneous than positive information, echoing Tolstoy’s famous adage that: “all happy families are alike; each unhappy family is unhappy in its own way.” Many studies of “information ecology” (19,20) now support Tolstoy’s claim, showing that there are more diverse ways of communicating negative (versus positive) information about emotion (21,22) and describing other people (23,24). Second, negative information is more counternormative than positive information because positive comments are more common than negative ones (25), which makes negative comments stand out (26). Because negativity is heterogeneous and counter-normative, users may unintentionally write negative comments from a desire to stand out in crowded online spaces. [...]
Current Research:
[...] Our observational analysis draws from 2.05 billion comments from 2,150 topic-specific “subreddit” communities on Reddit across 2010 - 2022. [...] We then run a large-scale experiment in which 4,000 participants post comments across five “generations” of an evolving social media feed. We manipulate the motivation to be unique by either incentivizing participants to either contribute unique content or to conform to prior content (i.e., disincentivizing the motive to be unique). [...]
[...]
Discussion (findings):
Across 2,150 Reddit communities including 2.05 billion comments, and in a large multigenerational experiment, comments on social media became more negative over time. Three patterns in our data suggest that this trend arose at least partly as a byproduct of users’ motivation to differentiate themselves.
First, the trend towards negativity was explained by semantic differentiation. Using a well-established deep learning method of capturing semantic similarities and differentiation (31), we found that negative comments had higher levels of semantic differentiation, and that controlling for semantic differentiation statistically either partially or fully explained the trend in negativity over time, depending on the exact analysis. In our experimental dataset, we manipulated participants’ motivation to either differentiate themselves or to conform to norms through monetary incentives. We only found a trend towards negativity when participants were incentivized to differentiate themselves.
Second, the trend towards negativity was moderated by the positivity of dialogue. It was steepest when conversations were highly positive, but it flattened or reversed when conversations became more negative. This flattening was not because people became less likely to leave negative comments (a classic “floor effect” explanation). Rather, it was because the rise in negative comments was matched with a rise of positive comments. We suggest that, as discourse becomes more negative, both negative and positive comments allow users to differentiate themselves: positive comments are unique in negative threads because they are counter-normative, whereas negative comments can be unique because negative information is heterogeneous; there are many unique ways to express negativity.
Third, we found that group size showed the same effects of time in our observational dataset. We view group size as theoretically similar to time. Just as users later in a discussion have more pressure to differentiate themselves than users earlier in a discussion, users in larger communities have more pressure to differentiate themselves than users in smaller communities. Consistent with this prediction, larger communities were more negative than smaller communities, changes in community size were associated with shifts towards negative dialogue, and both of these effects were statistically explained by semantic uniqueness. Our analyses disentangle any covariation between community size and age to ensure that these are independent predictors of negativity, rather than confounded predictors. [...]
Our supplementary materials also include a range of other analyses, such as testing whether these effects characterize both political and apolitical social media communities, and testing whether the shift towards negativity also characterizes online news sites. We show that our effects generalize across both kinds of social media community, but do not generalize to the news. This is consistent with our theory, since the news does not meet our key theoretical criteria (people making sequential comments on the same topic in an evolving discussion thread). The fact that our effect generalizes across political and apolitical communities is important, because it suggests that motivated differentiation is a fundamental mechanism that can make social media dialogue more negative even when intergroup competition is not salient.
One interesting feature of our observational results was that dialogue never became highly negative at the aggregate level, even after valence declined over time. Rather than crossing the neutral “0” point of valence, we found an equilibrium around 0.20. In our view, this equilibrium reflects the fact that human perceptions of neutrality are seldom aligned with true neutrality. Negative information is more dominant than positive information, which means that dialogue must be more extremely positive to be subjectively experienced as positive, whereas dialogue can be mildly negative to be subjectively experienced as negative (26,32). Comments like “Sure, I guess you’re right” and “I’m fine” may likewise receive mildly positive scores in a sentiment classifier such as VADER, but would be perceived as negative by users. [...]
The most troubling aspect of our findings is that they suggest that negativity on social media could be relatively robust to intervention. Intergroup animosity may improve when algorithms explicitly promote content that “bridges” the preferences of liberal and conservative users (33,34), and algorithms that promote prosocial content may alleviate users’ misperceptions that moral outrage is normative in online social networks (35). However, it is more difficult to disincentive the motive to be unique on social media. Furthermore, our research suggests that algorithms that promote positive and prosocial content may even have unintended backlashes because it incentivizes negative dialogue by making it even more unique. Differentiation-based negativity may prove to be a fundamental obstacle to sustaining positive dialogue on social media. [...]
Edit: So I suppose the question now is how can we differentiate ourselves without resorting to negativity.
Sounds interesting. I wonder how we should take into account upvotes and likes? Because positive comments are often not that differentiated, we have standard UI's for expressing positive sentiment...
Sounds interesting. I wonder how we should take into account upvotes and likes? Because positive comments are often not that differentiated, we have standard UI's for expressing positive sentiment and often don't post that sort of thing as a comment at all?
Maybe letting people upvote posts using emojis isn't such a bad thing?
It might be statistically significant but it's pretty much nothing. The valence changes are also ridiculously small.
It might be statistically significant but it's pretty much nothing. The valence changes are also ridiculously small.
Among the first 5 comments in each thread, 25.21% were classified as negative, compared to 25.96% among the final 5 comments—a modest increase of 0.75%. Although this difference is small, it is highly robust. When aggregating comments by their sequential position, the proportion of negative comments strongly increased over time
On one hand, it seems right that the effect should be small. Most meme threads tend to remain meme-y and rarely get to the point of needing to be locked or something. One the other hand, a small...
On one hand, it seems right that the effect should be small. Most meme threads tend to remain meme-y and rarely get to the point of needing to be locked or something. One the other hand, a small measurable difference can still feel big due to negativity bias, and I think this is one of the areas where perception matters as much. Would be interesting if there's another study comparing how this valence indicator correlates with how the participants reported to actually feel.
That's a fascinating study, with sobering conclusions. I hadn't thought of it as a matter of wanting to stand out. Perhaps that's part of the enduring appeal and continuous churn of counterculture groups? What it says to me is that there's benefit in having spaces outside of the mainstream where people can go and express their differences purely for the feeling of being different, even aside from any actual differences.
My solution: We need more (benign) cults, secret societies, hidden libraries, and other forms of non-toxic differentiators and social groups that don't harm the people involved, which can inoculate them from the appeal of anti-establishment groups that are more harmful to society and to the people involved.
Your suggestions all sounds like a lot of fun actually, just imagining them makes me feel like wanting to find and join them, I think you're on to something here
Anecdotally, my experience is communities that focus on working toward a goal or completing a project also tend to be more supportive. Maybe something about being part of a higher cause softens the desire to stand out
Fan fiction and conspiracy as well. The appeal of something made up that we can invest in and be unique easily. Expend our extra mental cycles. Idle hands.
Even when I was forming this post I searched for 'conspiracy' in the comments to see if it was novel. In this case I would have self-censored had someone mentioned something different. But my first thought was self-censor, not negativity.
Man this study is awesome and it brushes up against some of my own personal hyperfixations such as online propaganda and online tribalism.
I think there's an even deeper psychological connection here between differentiation, identity, perceived superiority, and hostility.
There's a concept in social psychology called "positive distinctiveness," which describes how people don't necessarily just want to be different from others, they want those differences to reflect positively on themselves and/or their in-groups. People generally want to view their identities, beliefs, and social groups favorably, and one way of accomplishing that is through another concept called "social comparison."
And one of the easiest ways to establish a sense of superiority isn't necessarily to demonstrate your own intelligence, morality, or competence, but to portray someone else as less intelligent, less moral, or less competent.
That's where I think this study gets particularly interesting.
People increasingly tie their opinions to their identities. An opinion stops being something they believe and becomes part of who they are. Consequently, someone challenging that opinion isn't necessarily perceived as presenting an alternative perspective, but as threatening their identity and their positive perception of themselves.
This connects to the social psychology concepts of "identity-protective cognition" and "motivated reasoning," where people become more inclined to defend information that reinforces their identity and reject information that threatens it, regardless of its factual accuracy.
Now combine that with the desire for differentiation demonstrated in this study.
People want to stand out, but they also want to feel good about what makes them different. Negativity offers an accessible way to accomplish both. Criticizing someone else's opinion can simultaneously distinguish your contribution, establish perceived intellectual or moral superiority, and reinforce your identity within a social group.
And social media can reward this behavior through attention, engagement, and validation from like-minded people.
There's also a fascinating connection to Optimal Distinctiveness Theory, which proposes that humans have competing needs for individuality and belonging. Online tribalism can satisfy both simultaneously. You establish your individuality by identifying with a group whose beliefs distinguish you from outsiders, while gaining belonging and validation from people who share those beliefs. While conflict with others in the outside groups reinforces those beliefs.
My hyperfixation on online propaganda is actually what led me to explore these psychological mechanisms in the first place, because propaganda exploits precisely the motivations and vulnerabilities we're discussing.
I have found a particularly effective propaganda strategy isn't simply convincing someone that a particular claim is true. It's getting them to incorporate that claim into their identity, especially by associating it with being smarter, more informed, more moral, or more enlightened than people who disagree.
What this means is that a lot of disinformation and propaganda isn't just framed as "information", it is ALWAYS spun with emotionally charged language and frames it in the context of "in-groups" and "out-groups", even if those groups are just ideologies and not specific groups.
Once that happens, contradictory evidence can become psychologically threatening. Accepting that you might be wrong no longer means merely reconsidering an opinion. It can mean questioning something you've used to define your identity, your social belonging, and potentially your sense of superiority.
And when attacking opposing viewpoints earns social validation, the hostility itself can become reinforcing.
What's especially interesting is that these mechanisms don't necessarily require people to be consciously seeking superiority, and they can operate across ideological and political boundaries.
Obviously, this study doesn't establish that entire process. It specifically demonstrates a connection between the motivation to differentiate and increasing negativity. But when you combine its findings with existing research on social identity, positive distinctiveness, social comparison, motivated reasoning, and status-seeking behavior, I think there's a compelling larger hypothesis worth investigating.
Social media may not simply amplify existing tribalism and hostility. Its incentives for differentiation and social validation may actively encourage people to build identities around increasingly oppositional beliefs, then reward them for defending those identities through antagonism.
And that creates an environment in which propaganda can potentially become especially effective, because people aren't just defending what they believe. They're defending who they believe themselves to be.
Woah I haven't thought of how all the pieces fits together like this, thanks. I know personally it's quite difficult to take someone critquing an argument I agree with not as also an attack on my belief, the dark pattern is hard to avoid.
Link to preprint (free to read version). Excerpts:
Introduction:
[...]
Discussion (findings):
Edit: So I suppose the question now is how can we differentiate ourselves without resorting to negativity.
Sounds interesting. I wonder how we should take into account upvotes and likes? Because positive comments are often not that differentiated, we have standard UI's for expressing positive sentiment and often don't post that sort of thing as a comment at all?
Maybe letting people upvote posts using emojis isn't such a bad thing?
I guess we'll need some more differenciation to reaffirm the positive affect...
I was ready to assume you made a typo, but something about it nagged at me... Is this what you're referencing?
Yup, bingo. It was just meant as a pun for my own entertainment (and posted for the chance passersby), but I'm not entirely joking either.
Yeah I'm trying to parse their definitions and it's surprisingly apt, looks like there's a rabbit hole to fall into.
If you like metaphysics (and who doesn't?), Difference & Repetition is one of the more insightful books I've read.
I enjoyed the pun!
It might be statistically significant but it's pretty much nothing. The valence changes are also ridiculously small.
On one hand, it seems right that the effect should be small. Most meme threads tend to remain meme-y and rarely get to the point of needing to be locked or something. One the other hand, a small measurable difference can still feel big due to negativity bias, and I think this is one of the areas where perception matters as much. Would be interesting if there's another study comparing how this valence indicator correlates with how the participants reported to actually feel.