So how might it be improved upon to form a better model? The first step is so obvious that people quite often forget it is a step at all, and that is to make the 0--6 range a continuum, rather than seven discrete points. We can now describe our sexuality as a point in a one-dimensional space. (If the term "one-dimensional space" worries you, stop reading now because the maths is going to get worse.) So having got this concept down, let's take a step back and re-evaluate what we're dealing with.
What we have is essentially an axis which has attraction to MOTOS (Members Of The Opposite Sex) in one direction and attraction to MOTSS (... Same ...) in the other. These two aren't mutually exclusive, so why don't we have two axes, one for MOTOS and one for MOTSS, and set them normal to each other? This results in a description of sexuality which uses a two- dimensional space, with a person's sexuality being represented by a point, one ordinate of which measures their attraction to MOTOS and the other their attraction to MOTSS (Figure 1. below). Now, suppose that we say that point E represents someone who is entirely homosexual, and point A someone who is entirely heterosexual. The line joining these two points (ACE) is effectively a continuum Kinsey scale (given a suitable scale). So if we choose to describe our sexuality 2-space by axes at 45° to the original MOTOS and MOTSS axes (Figure 2.), one will be a Kinsey scale (x') and the other ... what? Consider the x axis on Figure 1. which measures amount of attraction to members of the same sex, so B is more attracted to MOTSS than C. And likewise with the y axis, B is more attracted to MOTOS than C. And C is more attracted to both than D (D not having any sexual interest at all). So y' runs from someone with no sexual attraction, through someone with some attraction to both MOTSS and MOTOS, to someone with more attraction to MOTSS and MOTOS. Which means we can interpret y' as being strength of sexual interest in people (the nature of the people they are interested in being determined by their x' co-ordinate), or to put it more crudely, sex drive.
If you make that inverted bell-curve (the distributions aren't actually important here) a heterosexual to homosexual continuum, instead of a male to female one, isn't that very much like what I've been describing? The lower diagram looks a lot like my Figure 2. with the x and y axes from Figure 1. drawn in, representing amount of male identity and amount of female identity. The "conventional" axes would be a horizontal (comparable to x') one for the usual male--female axis, and a vertical one (comparable to y') representing ... what? This time, I don't have much of an answer. By analogy, it is "intensity of sexual identity", but this isn't a concept I've heard anyone talk about before. It appeals to me, but maybe I've made a mathematical construct which has no meaning in the real world?
Left to themselves, most people conceive of male and female as two separate spheres, with no intersection on the Venn diagram. But when someone comes up who does not fit this worldview -- like Sparrow -- they change the binary opposition into a continuum, forming an inverted bell curve:
The left side of this curve they call male. The right side they call female. The middle they call terra incognita, and then they draw in sea monsters and locate the kingdom of Prester John.
With some pairs of opposites, like dark and light, it makes perfect sense to speak of a continuum. Dark is the absence of light; hence all mixtures of dark and light are much the same, whether we call them dawn or twilight. But male is not the absence of female, nor vice versa; both are presences, which we choose to regard as opposed. So there is a difference between a person who is both male and female, and a person who is neither. We should really plot gender like this: