That's true, if it's something simple (loads of forms can't possibly need it that badly - like surveys, they just want it for statistics) they can just lump all the "other"s into a group on the pie chart. Or if, say, half of all people who ticked "other" wrote "genderqueer" they could make it a group.
Yeah, it would be interesting to see what people wrote down, but if they said their gender was toast, they're probably not putting (the right kind of) thought into their other answers either!
Yep, it's often used to make sure the interview process isn't biased one way or another. Like if you had 60% men and 40% women apply for a job, but 90% of jobs went to women, there's something weird happening along the way.
For pharmacy and stuff, it is important to have that kind of data. For example, some drugs are dangerous to people who might become pregnant, which includes no biological males. But it still doesn't work because then trans people have to either put the wrong thing on the form (ie their biological sex when it asks for gender) or, depending on where they are in a transition process, might end up with the wrong dose, or counted in the wrong survey group.
So, maybe for surveys and such it should be
( ) Male
( ) Female
( ) Other _____________
and for medical things, or things that are for studies involving chromosomes, hormones, etc, it should be
( ) Biologically Male
( ) Bioogically Female
( ) Transsexual Male to Female
( ) Transsexual Female to Male
or some variant? Then genderqueer etc people would answer on their biological sex and trans people would be included correctly. (Not perfectly, since everyone's different and particularly since people would be at different stages in transition, or choosing different paths for transition, etc, but at least they could be accounted for. And if it's for giving them a drug or something the doctor would know to check.)