I think that his point is that the "why" question doesn't fit well.

There is no purpose to any mutation. They just happen.

Some of them, if they're disastrous or unlucky enough, get selected out over time. But even very bad ones can linger pretty much indefinitely if they often lie unexpressed for generations at a time.

The more advantageous ones will have a higher chance of increasing, but being disadvantageous is no garauntee of going away.

Consider blue eyes. They are a bit disadvantagous vision-wise, which hurt their chances of survival over the long term. But they're also recessive, so not everyone with a gene for blue eyes will have blue eyes, which confers a significant buffer between blue eyes and natrual selection. Thus, you might expect that while blue eyes do keep happening, they would be noticeably rarer than brown eyes. Is that the case?

So, suppose that homosexuality were entirely genetic. Obviously, it is extremely disadvantageous when it comes to reproduction. But it is not a terribly dominant genetic strain, given how often the children of gay people are not gay, and the proportion of gay people born to straight parents. So what would you expect to happen? Much like the case of blue eyes, you'd expect it to be rare, but still bubble up.

Evolution is an extremely sloppy process with lots and lots of randomness. Mutations do not have to be at all reproductively beneficial to survive; being so just gives them better odds at being more wide-spread.