[TomCowley]

Sickle-cell anemia is a simple example of how a "bad" gene can persist. 2 copies are clearly bad, but 1 copy confers a degree of resistance to malaria. The gene actively kills some people and stops them from reproducing, but it actively helps others reproduce, so it survives.

Note: The next paragraph is pure speculation to provide an example- I have no evidence that the genetic basis of homosexuality fits this model.

Since the genetic component of homosexuality is definitely more complicated (and studies of identical twins separated at birth confirm a genetic component), let's assume we have multiple genes that come together to give a degree of predisposition toward homosexuality. If each of these genes, in the absence of the others (or the absence of enough of the others) confers an advantage, then just like the sickle cell gene, it will propagate and multiply in spite of the fact it directly causes some offspring to be unable to reproduce.

If there were exactly 1 gene for homosexuality, and it had no other effect besides a gay/straight switch, you would be correct that the trait is most likely not old (it could still get lucky and propagate for awhile), but homosexuality is obviously not the sole result of one gene with no other side effects, so your conclusion does not follow, as the two examples above show.