The philosophy of religion of the last 30 years or so has done a good job of showing that the argument from evil has to rest on gratuitous evil, but there is no consensus or even much inquiry over what exactly such an evil is. Sure, gratuitous evils are somehow pointless or not leading to some greater good, but this can happen in at least four ways. Evils might be justified as ordered to either (a1) an actual good or (a2) a possible good and the evils themselves can be considered either (b1) in particular or (b2) in general.
(a1)(b1) If evil has to be justified in this way, then every particular evil has to be seen as ordered to actual greater good. If we can’t see any concrete good greater arising from the fawn that’s burned in the forest fire or little Timmy dying of cancer, then the evil is gratuitous. If we further stipulate that the good that has to arise has to be peculiar to the evil in question and that it can’t be eschatological or far remote, g-evils seem plausible.
a1b2 If evil has to be justified in this way, then there must be some actual good that arises out of an evil, but only so far as it is a sort of thing. On an account like this, we might say that our bad backs were a consequence of bad luck, and the bad luck was allowed so that backs could be generated by natural processes like selection. This does not give a justification for bad backs as such, but only a justification for allowing things to happen by selection. Still, there is a concrete good that arises from allowing evils like bad luck, sc. the efficacy of secondary causes. If a g-evil is one that lacks a justification like this, they seem a good deal less plausible.
(a2)(b1) On this account of evils, every particular evil is justified because it makes some greater good possible, but there is no necessity that this greater good ever come about. So maybe the point of my back pain is to make heroic charity possible or inspire others to flee from complaint at my bad example, even though I never develop any charity and no one else leaves off bellyaching. It’s unlikely that all evils could be justified in this way, but this does not preclude an indefinite number of them being so. On this account, the justification of evil opens up to counterfactual possibilities which would be remarkably hard to calculate even in a single case, especially once evils start causing each other.
(a2)(b2) On this account, evils are justified only in general, and an indefinite number of them might lead to no actual greater good at all. So perhaps the fawn that burns in the woods is only justified as an instance of animal pain, but animal pain existed only for the sake of a good which, while making it possible, never came to pass. The amount of counterfactuals now become incalculable and the parameters of possible goods are probably unknowable.
These justifications are orthogonal and all or none of them might play a role in a possible theodicy. If we assume all justifications have to be type-1, the burden on theodicy is very high and we might even let Rowe get away with assuming that it’s obvious that there are g-evils.