Vector identity theory
In my big news post, I hinted at a post about "vector identity theory" aka "I am a giant geek". Here's some stuff about how I think about identity groups and labels.
By "identity labels", the core of what I mean are nouns that you use to describe people -- "woman", "man", "goth", "punk", etc. However, many more "adjectival" things fit too -- "masculine", "feminine", "trans", "queer" (oh god please don't call anyone "a trans" or "a queer" -- it just sounds wrong). In any case, these things are useful for communication. They allow people to tell others some shorthand about what their life is like. They allow people to find more people like themselves to compare notes with. I use them a lot.
The problem is that they're wrong. Or, rather, not quite right. Any time you have an identity, it comes with a pile of stereotyped behaviors that any given claimant of the identity might or might not share, and it tends to reduce the perception of the claimant down to those stereotypes. Oops.
Now for the geeking out (note: I'm mostly just playing here).
I (sometimes) view identities as vectors in some huge or infinite-dimensional space. Given a set of identities (say, {male, female} or {straight, queer} or {gay, lesbian, bi, trans, queer, questioning, ally} or whatever, finding out how you identify is a process akin to estimating the projection of your personal self-vector onto the subspace covered by the basis of identities in the set. Of course, that basis is never orthonormal; that would be too clean. It's not orthogonal or normal at all. It's just a mess of huge-dimensional vectors that you have to try to match yourself up against, throwing away all those components of yourself that aren't in directions available to you in that basis. Worse, the self-vector is a function of time. The way you projected last time might be a good predictor for this time, but then again, it might not. Even the identity labels change over time. Does being a goth mean the same thing now as it did fifteen years ago?
All that assumes you have some continuously-variable scale you can use to communicate your coefficients, even if it is in qualitative words like "very goth" or "a little bit femme", and that you can assign coefficients to each identity in the space more or less independently. Then you get situations where people ask things like "are you a man or a woman"? Two problems! First, "man" and "woman" don't seem to me to be completely dependent identities -- they have somewhat different directions to me, not just one in the opposite direction of the other -- so you would at least need two coefficients to answer that questions. Second, the asker usually expects not only a single coefficient, but a single coefficient of -1 or 1. How absurd! I've probably thought about this too much, but I'm left somewhat surprised at how anyone ever has a useful answer to questions like that.
By "identity labels", the core of what I mean are nouns that you use to describe people -- "woman", "man", "goth", "punk", etc. However, many more "adjectival" things fit too -- "masculine", "feminine", "trans", "queer" (oh god please don't call anyone "a trans" or "a queer" -- it just sounds wrong). In any case, these things are useful for communication. They allow people to tell others some shorthand about what their life is like. They allow people to find more people like themselves to compare notes with. I use them a lot.
The problem is that they're wrong. Or, rather, not quite right. Any time you have an identity, it comes with a pile of stereotyped behaviors that any given claimant of the identity might or might not share, and it tends to reduce the perception of the claimant down to those stereotypes. Oops.
Now for the geeking out (note: I'm mostly just playing here).
I (sometimes) view identities as vectors in some huge or infinite-dimensional space. Given a set of identities (say, {male, female} or {straight, queer} or {gay, lesbian, bi, trans, queer, questioning, ally} or whatever, finding out how you identify is a process akin to estimating the projection of your personal self-vector onto the subspace covered by the basis of identities in the set. Of course, that basis is never orthonormal; that would be too clean. It's not orthogonal or normal at all. It's just a mess of huge-dimensional vectors that you have to try to match yourself up against, throwing away all those components of yourself that aren't in directions available to you in that basis. Worse, the self-vector is a function of time. The way you projected last time might be a good predictor for this time, but then again, it might not. Even the identity labels change over time. Does being a goth mean the same thing now as it did fifteen years ago?
All that assumes you have some continuously-variable scale you can use to communicate your coefficients, even if it is in qualitative words like "very goth" or "a little bit femme", and that you can assign coefficients to each identity in the space more or less independently. Then you get situations where people ask things like "are you a man or a woman"? Two problems! First, "man" and "woman" don't seem to me to be completely dependent identities -- they have somewhat different directions to me, not just one in the opposite direction of the other -- so you would at least need two coefficients to answer that questions. Second, the asker usually expects not only a single coefficient, but a single coefficient of -1 or 1. How absurd! I've probably thought about this too much, but I'm left somewhat surprised at how anyone ever has a useful answer to questions like that.