Thoughts on identity, personhood, and social media

[mangosteen]

Context: What follows is a thought that I need to develop more, but the nucleus was worth posting.

1. Facebook as a company constantly talks about making connections between people; but it can't, and it doesn't. Rather, Facebook makes connections between identities, because that's the most immediate thing it has access to: relationships (e.g. who you friend, who friends you, who you block/mute/etc.), and affinities in the forms of likes, shares, and postings.

2a. Facebook's problem is an interesting one: Identities don't click on ads and buy products. People do. To wit, someone who creates a Facebook account as a way to express one particular aspect of their personality will give a very skewed view of their ad preferences, and likely won't see relevant ads. Conversely, the more aspects of a person's personality are expressed in their postings on Facebook, the more relevant the ads are going to be.

2b. Since advertising is a numbers game, you as an individual aren't interesting, but "everyone like you" very much is. Ergo, a collection of interlinked people is a very potent engine to create the most valuable thing for advertisers... the largest pool possible of likely customers.

3. Facebook's goal is therefore to make people and identities as much of a one-to-one mapping as possible. One global person = one global identity = a stable target for ads.

4. Okay, so how do you do you make "identity" and "personhood" converge, given that humans don't really do "being the same person to everyone" very well? Well, we can look at what Facebook has already done to see how they are solving it: Facebook Login, Facebook Pixels on websites, The "Real Name" policy, making it more difficult than necessary to limit information, etc... but that all feels like begging the question a bit.

5a. To do this from first principles, the better question to ask first is "How are people and identities different?".

5b. The next question is "Who is identity convergence useful enough to, to pay real money for the service?", because that shapes the resources available.*

That's where I am right now.

---

* This is also the reason that Facebook is free for users. A user couldn't/wouldn't pay enough to make up for their potential lost revenue, presuming that making people pay will shrink the pool of users. See: advertising is a numbers game.

Comments

[jducoeur]

Suffice it to say, I've spent *vast* amounts of time thinking about this one for Querki, albeit from the opposite perspective of Facebook.

Querki actually has three distinct concepts of person-hood built in:

-- The User represents an actual logged-in human being, who may be a member of any number of Querki Spaces.
-- An Identity is a distinct *view* of a User, generally based on some public identity somewhere.
-- A Person is the representation of an Identity's presence in a specific Space, which can be customized to that Space.

The important bit is that the Identity-to-User mapping (and thus, the mapping between Identities) should not be visible to anybody except the User unless they *very* explicitly opt into it. This is so that you can potentially make use of social networks from multiple sources, but by default nobody should be able to tell whether this person from Facebook, who is a member of Space A, is the same as that person from Google, who is a member of Space B, unless you specifically *want* to cross the electron streams.

It's surprisingly challenging to get right. I've put man-months of effort into it, and I suspect I'm still going to find a few distressing bugs when it finally becomes relevant. (Which will be when I turn on OAuth integration.) I'm necessarily being explicit that this is best-effort: Querki is trying hard to get it right, but I don't want anybody for whom identity separation is life-or-death important depending on it.

It's worth noting that the only reason this is even relevant is user convenience: being able to use all of your social networks, within a single login session, without sacrificing privacy. It'll be interesting to see whether it was even remotely worth the effort, but if nothing else, it's been damned educational.

I'm not yet convinced that it's possible to provide the level of privacy guarantees I *want* to be able to give; if I ever puzzle that out to my satisfaction, I suspect it'll be worth separating it out into a third-party service. I'm tempted to do so on general principles, but it's enough effort that I can't afford to just do so on a lark. And I suspect that, as a separate service, it would be painting a huge target on its back from the big players -- as it is, I half-suspect them to prevent me from making it work as designed...

[tamidon]

as a non-techy, when I first joined the online world someone told me "If you don't pay for the service, you are the product". Aside from that, you think about this stuff way more then most people I know.

[gale_storm]

It was relatively recently that I first came upon the term of the Facebook pixel, and then I started with the thought of how could it happen to trick them into believing it or counting on it or ... gah! Okay, it isn't something that is easily attenuated, but perhaps that's just part of my identity.

[drwex]

Just an FYI for your calculations, FB's latest SEC filing shows that its cost-per-user is REMARKABLY small. I forget the exact figure, but it's about a buck.