[Aloisius]

> Does that make a highly accurate approximation into a less accurate one?

What we have right now is not an approximation, but a binary measurement. Turning sex into an approximation would be considered a significant step forward considering science tells us it is not binary.

[Kalium]

As simon_ and Symmetry point out, that binary is in fact a pretty accurate approximation. So it seems we have both a binary measurement and an approximation in one. Perhaps we should consider that something can be both a binary measurement and a reasonably accurate approximation of a spectrum.

Regardless, my earlier question stands. At what point is it acceptable for an approximation to not be accurate in detail? What's the acceptable level of error in approximations?

[TuringTest]

> What's the acceptable level of error in approximations?

I'm fairly sure the lifes of those affected negatively by the error are outside the "acceptable level". We're talking about humans here, not mathematical rounding errors.

[Kalium]

It seems your objection is not to approximations or the error inherent in them, but to what happens when people forget approximations are not reality. Is that correct?

[nickff]

The 'two-sexes approximation' is boolean, not binary. Boolean involves two possible values; binary involves an infinite number of possible values, represented in base-2.[1][2]

[1] http://en.wikipedia.org/wiki/Boolean

[2] http://en.wikipedia.org/wiki/Binary

[quadratical]

Not to be pedantic, but binary is entirely correct in this context. We aren't talking about computer science. Binary means relating to, composed of, or involving two things.

Boolean is simply a binary variable, having two possible values: true and false.