Why getting identity expression wrong is dangerous for social networks
“You have one identity … The days of you having a [single] image for your work friends or co-workers, and [another] for the other people you know are probably coming to an end pretty quickly … Having two identities for yourself is an example of a lack of integrity.” - Mark Zuckerberg, making a terrible mistake
When I was younger I remember being frustrated that I didn’t know ‘who I was’. As a creative attempt to solve this problem, I decided to write down everything I knew about myself. It ended up being a somewhat random collection of things along the lines of ‘likes Ariana Grande’ and ‘English Literature grad’. After I had finished the exercise I remember feeling a little unsatisfied. These were things that were (unfortunately) true about me, but they didn’t feel like they were me.
A long time and a productive shrooms trip later (disclaimer for future employers: this is a metaphor), I came to realise there is no ‘me’. This isn’t a The Sixth Sense style ‘gotcha’ where I reveal to you this post was written by GPT-3 - what I mean by this is I realised trying to pin down a fixed idea of ‘me’ wasn’t helpful. This previous idea I held about the self being some collection of things was actually quite limiting. At any point in time it’s true I might possess some number of traits and behaviours which any person might accurately ascribe to me, but I noticed that these things were all fluid depending on the context I was in and more importantly the decisions I was making (hold your free will challenges for another post please). Thinking of the self in this way trapped me into doing certain things even if they weren’t valuable to me, because I had identified as the sort of person who did those things.
Evolutionary psychology actually supports this view of identity as fluid - the argument is that it’s something we use simply to facilitate and guide required social interactions. Yet, despite identity playing a central role in almost all consumer social apps, the limits and purpose of identity hasn’t really been applied as a lens to the industry. And given it appears to exist for the sake of social cohesion, catering to how identity actually works is really important if you want to design products that influence social interaction in a helpful and healthy way.
What is identity?
Paul Smaldino says identity basically helps humans to solve this problem: given some goal, how do I know who of the many available human beings I should work with, and how I should work with them? Identity signalling helps individuals find partners who have shared behavioural and communicative norms. Let’s say you need to form a hunting party but you live in a populous, diverse community where you don’t know everyone personally. It’d be a dangerous endeavour to go with a random group of people, taking a chance on their trustworthiness and how well you might work together. Using identity ‘signals’ as a proxy here is efficient. Maybe one identity group comes from a cultural background where people are more individualistic and another from a group where people are more collaborative. It's important to understand who matches your expectations before this theoretical hunt. Identity acts as a vessel for sharing and understanding expectations when it comes to cooperation with others. Turns out dressing up for Comic-Con isn’t actually about your favourite anime character at all - it’s about telling others what they can expect from your behaviour (in this case: that you’re a loser).
This means identity signals are actually much less important in smaller communities: if everyone knows everyone then you can rely on simply knowing the people around you as individuals instead of trying to guess their behaviours based on a signal. There’s a super interesting study which investigated how people perceived the faces of potential sexual partners across differently sized populations. It found that the degree of preference for ‘masculine' faces in men (e.g. square jawlines) was correlated with the size of the social environment. People in smaller communities didn’t need to rely on ‘masculine' faces (which is actually a predictor of male aggression) to identify men who were aggressive (which is apparently correlated with attraction) - they could know each man by their reputation.
The opposite is true for larger societies, but what complicates things isn’t just the number of people. Even if the number of people in a community stays the same, adding additional social contexts in which people need to cooperate with others increases the number of identity signals that might be useful. Imagine a community where monogamy exists, versus another where it doesn’t. For single people in the former community, suddenly certain signals (e.g. a wedding ring) are necessary for the efficient sorting of people, that might be less important or not needed at all in the latter community.
Similarly identity signals are needed more-so when there is a greater risk of people not sharing the norms associated with that identity. If you’re a muslim in a muslim majority country it is less important to signal being ‘muslim’ to others, because you don’t really need to sort between those who share the norms of that identity and those who don’t. This is of course why no-one ever makes disliking James Corden a part of their identity - it’s a given.
The funny thing is this goes some way towards explaining the behaviour of the straight white men who want to outwardly signal and take pride in these traits as their countries become more diverse - it does fit with what this evolutionary model says about how you might expect someone in such a situation to react.
So turns out my (metaphorical) shrooms trip produced a genuine insight: identity is not some fixed thing and is in fact context dependent and deployed with a specific purpose. In an environment with diverse religious identities and uniform political identities, your religious identity is the more important thing to signal, and the thing you will more strongly identify with. But in an environment with diverse political identities and more uniform religious identities, the political identity becomes more important.
This explains identity as an outward expression - but we feel our identities inside too and seem to want to live in accordance with the rules and values espoused by them. So what’s going on here? You might have heard the Robert Trivers argument that from an evolutionary perspective it is advantageous for animals to believe the lies they tell to others, given this makes their lies more convincing. If you believe a lie you’re telling yourself, there’s no need to actively mask your own feelings and behaviours when trying to fool someone (e.g. to consciously avoid cartoonishly shifting your eyes from side to side). Burton Voorhees (et al.) applies somewhat similar logic to internalisation of an identity. If a person deeply believes in their identity, it’ll be clearer for others who are similar to identify that person as a good candidate for cooperation.
What happens to identity on the internet?
With this context, we can start to understand why things get a little bit crazy on the internet - and we can think through some ideas that might help restore calm and order.
Firstly the internet is the biggest ‘society' the world has ever seen - there are about 4.6 billion people on the internet (though of course not all are on the same internet given the Great Firewall). Regardless, it's big. Plus the internet has introduced many new contexts in which people need to interact. Think about gaming for example. In many multiplayer team deathmatch style games (e.g. Call of Duty), players come together and share the goal of beating other teams in a social context which didn’t exist in a comparable way prior to the internet. The social context is one where usually the bond between players lasts for a single game only (after which players are split up and rematched with others for the next game), where knowing the skill level of each teammate and enemy is particularly helpful, and where you have access only to the custom avatars of other players and their voices. In this context many traditional markers that would be used for cooperation (e.g. social class, muscle, etc) vanish and are replaced by new identity markers like in-game status symbols (e.g. a user’s level or special items). This is just one example - but the consequence of scaling up the number of people you might interact with, and contexts in which you might need to interact with them, in this way is ultimately a scaling up of the number of identities and associated signals that are used for sorting people.
Similarly the risk of people not sharing the norms associated with your identity on the internet is very high - there is less of a sense of a ‘majority’ identity - and so all aspects of your identity become more important to signal. As an Indian person in India you might not feel the need to signal being Indian as much in real life. However arriving on the internet where this is no longer the default, the theory above suggests this person would exaggerate their signalling to deal with no longer being the majority. You could argue that curated feeds should have the opposite impact - but I would argue that people retain a sense of the wider community of a platform even within a niche (e.g. if you read the comments TikTok users are clearly aware of the various niches to which they don’t belong on the platform). Similarly there is evidence to suggest echo chambers might not really exist in the way some would have us believe.
The net outcome here is the internet multiplies the number of social identities that people need, and the intensity with which they rely on and signal their various identities.
This has very real consequences. The expectations of co-operation that come with identity are also known by another name: culture. These ‘rules’ are the lens through which a person interprets everything and importantly this includes interpretations of physiological stimulation - or what we call ‘feelings’. Lisa Feldman Barrett describes the idea that physiologically all we experience is pleasantness or unpleasantness, and arousal or calmness. Then what we know as ‘feelings’ are just ways that we learn to interpret these inputs. You might feel unpleasantness and high arousal, but whether you’re scared or anguished is a label you’ve been culturally taught and are trying to apply to these sensations. Burton Voorhees (et al.) argues that successful cultures teach feelings that incentivise the punishment of deviation from the rules of said culture (e.g. ‘shame’) - which makes some sense, given a culture which does this is more likely to be preserved and maintained. This is important because it means exposure to signals of someone belonging to an unfamiliar social identity can trigger feelings like indignation and anger. Well now we have a problem - as we’ve just covered, there are more unfamiliar identities than ever before being generated by the scale of the internet, and everyone has a stronger desire to signal these identities. Uh oh.
The number of identities and associated signals increasing in larger, more complex societies actually has some useful purpose for the individual. It gives people more tools in their arsenal to adopt in order to identify people they’ll cooperate well with, across multiple different contexts. You speak and act differently around your university friends compared to your work friends and this is because you understand the differing context. You switch to signalling a different identity (that is no less ‘you’) in order to harmonise as best as possible with others for that given context. Facebook made the error of not accounting for this by giving you one friend list, one profile and a centralised feed. By not accounting for context it became a struggle for creators of (social) content to identity signal in a way that would maintain social harmony. If you used the signals reserved for university friends in your posts you would risk alienating your work network or family, and identity signals aren’t built to work for all contexts.
So the setup of the internet in many ways has increased the propensity for people to feel identity threat, and upped the friction when it comes to maintaining multiple harmonious identities. However, all is not lost! A lot of these circumstances are the result of product decisions and need not be the default state.
What can the consumer social tech industry do about this?
The first big lesson is context really matters for social interaction. One solution here is to bring back private contexts, which can happen in two ways. The first direction would be to have varying feeds to which a user might broadcast, each for specific contexts (e.g. Close Friends on Instagram Stories). The second direction is to have closed groups or communities (e.g. a WhatsApp group chat) in which content is shared. This would ensure identity signalling can be relatively harmonious for the individual and untainted by mixing contexts. Smaller communities will contain less noise due to a reduced need to intensify identity signalling and respond to identity threat. If Instagram decides to pivot back to social at some point they might benefit from allowing users to post stories to various lists, instead of just a homogenous list of ‘close friends’.
Alternatively we can design products that adapt to the number of social contexts people need to operate on. The idea of a ‘profile’ itself could be better aligned to how identity works. In apps where a user might need to operate in many different social contexts, one option is to encourage users to focus identity signalling in the content they share with private context-specific groups, as opposed to encouraging them to do this through a centralised profile. Other than Facebook a lot of consumer social apps seem to already do this - e.g. your profile on WhatsApp is literally just your name and a chosen profile photo. This minimises the amount of signalling that needs to happen through it, which works well when you need to be able to send a message to everyone from your plumber to your spouse to a work acquaintance. This allows users to decide on which identity signals are appropriate per piece of content they send, which allows for more fluidity and context switching because content is decided on a case by case basis.
The second option is to allow users to identity signal within their profiles, but have different profiles for different contexts. This is kind of the way it’s already playing out but through having different apps entirely for different contexts. LinkedIn is for the context of professional work, Snapchat is for the context of younger audiences connecting with their network of peers, Tinder is for the context of dating - and all of these offer free and in depth identity expression through their profiles in a way that doesn’t create social friction. The version of this that less companies are facilitating is to have different profiles within the same app for different contexts, which Gen Z already hacked their way to with finstas on Instagram a while back. There are MVP features of this on Instagram and Twitter etc which allow this in theory but they haven’t really leaned into it. Balaji takes this idea a step further by arguing that we will proliferate into a ‘pseudonymous economy’ where people can use avatars to mask their identities from all other contexts in which that identity might be problematic.
The last thing you can do is to reduce permanence in a product so that identity signals are temporary and therefore easier to deviate from. This has of course already been employed to great success by Snapchat and Instagram Stories but there are many places it has yet to be applied. When Spotify and Netflix inevitably introduce social components, they would probably want to anchor these in what you’re watching / listening to now (which is transient) versus your liked media (which is permanent). And the reason is you don’t want to end up in a Facebook-esque situation where a person’s likes are 10 years old and they’re no longer as big of a fan of Ugly Betty as their profile might suggest.
And they all lived happily ever after
I’m conscious that there are a lot of things here that still need to be resolved, which I haven’t offered a solution for. It feels like we would lose something if we regressed into private contexts entirely and didn’t peak outside of our defined identity groups. Plus leaning into ‘what’s the safest thing from an identity perspective’ might not always be the best thing - maybe there are more important things that come out of the size and diversity of the internet for which a greater number of more intensely signalled identity markers are a reasonable price to pay. However I’m hoping this framework for thinking about identity will become a useful mental model as we continue to try to solve for the ‘how do we all get along on the internet?’ problem. Maybe this framework will also help you to be more conscious about approaching your own identity from a less dogmatic lens, and to recognise where the culture is hijacking your monkey brain to signal an identity threat (which isn’t actually as much of a ‘real’ concern in modern times). And of course if the framework doesn’t end up helping with that, hit me up because I know something that might...