If we define harm as being to actual children (not religion/impure-thoughts kind of harm), then the only argument is that future children will be harmed, because the viewer of the porn will have his sexually deviant longings for children reinforced and then, one day, will rape/molest a real child.
Personally, I do not buy this argument at all. I think it is very likely that virtual child porn (comic, animation, CG) prevents molestation overall. Because honestly, what do men do when they watch porn? Mainly, jerk off. And then once that's done, the urge is substantially dissipated... and so the subject would be less likely to put themselves at risk committing a sex crime against an actual child (something I understand a lot of molesters also feel guilty about).
So my hunch is that virtual child porn is probably a net positive. But I doubt we will ever get good data about this, and anyway it is one of those issues where logic isn't so likely to prevail.
Except the concern isn't about satiating the pedophile's need; it's about the pedophile becoming desensitized and eventually normalized to the thought of having sex with a child. They experience "loli" or virtual children, and that would be the gateway to real ones.
If you're talking about a trained psychologist prescribing the material to a pedophile, that would be fine, but the vibe I'm getting from this discussion is a bunch of people think if you just open the loli flood-gates, all the pedophiles will stop raping children, and that just won't happen.
One possibility I'd offer: comics can be used to teach empathy. Erotica can portray things in unrealistic senses, but creative writers can and have constructed elaborate plots that explore these unrealisms and their potential consequences.
Heck, look what School Days did for harem-style anime.
Personally, I do not buy this argument at all. I think it is very likely that virtual child porn (comic, animation, CG) prevents molestation overall. Because honestly, what do men do when they watch porn? Mainly, jerk off. And then once that's done, the urge is substantially dissipated... and so the subject would be less likely to put themselves at risk committing a sex crime against an actual child (something I understand a lot of molesters also feel guilty about).
So my hunch is that virtual child porn is probably a net positive. But I doubt we will ever get good data about this, and anyway it is one of those issues where logic isn't so likely to prevail.