It's easy to exclude the moon, like the IAU does, by requiring that planets orbit the Sun. (Although you might want to avoid the IAU's parochialism, which excludes exoplanets, and say "a star" instead.)

> By that logic, if we want to consider pluto a planet then the earth's moon also has to qualify. I don't really care one way or the other but that just doesn't seem sensible to me.

Ganymede is larger than Mercury, so a definition that doesn't specify orbiting a star would have to include some moons, or exclude Mercury.

This exposes another problem with this kind of definition - it seems to make more sense to define a body in terms of its intrinsic properties rather than whatever location it happens to be in.

For example, a rogue planet is not a planet by the IAU definition. You'd have to call it something like a free-floating planetary-mass object. ("Free-floating" is not the right term either, but I don't have the research funds to come up with a better one.)

> I have something of a personal hatred of poorly executed retconning

Totally agree. Wanna join me outside George Lucas' house to protest the Kessel Run?

> orbits and rogue planets

I think that's an example of coming to a useful realization while working through the process. It should serve to illustrate the importance of not attempting to engineer a particular outcome. If the earth were wandering the galaxy clearly it should still be considered a planet, no? And were we living on a body the size of mars that orbited a gas giant we would still be said to occupy a planet. At least assuming the goal is to classify things of a vaguely earth-like nature.

I think the more difficult question is what exactly qualifies as a moon. Anything orbiting a planet, including a smaller one? That seems perfectly reasonable to me. (Yes yes I know the barycenter matters.)

And yes, I think ganemede should be considered both a planet and a moon. That thing is on the same order of magnitude as mars. What useful purpose is served by excluding it as a planet?