On the one hand there are genuine semantic disputes, where participants dispute as to whether which is the correct analysis of a given target term or concept (of the sort of Karen’s martini case we have already discussed here). These are cases “under the scope of conceptual analysis” where premise 2 fails (see p. 201): “use” fits one of the proposal better than the alternatives.
On the other hand there are merely apparent disputes, in a certain sense to be dismissed—what Siders calls ‘no-fact-of-the-matter’ cases. He seems to characterize them as cases where there is there is semantic indecision between the alternatives: nothing in “use” (nor in the “eligibility” of the options) settles one option as the semantic right one. I think I agree that this is a sufficient condition for (true) “dismissivism” (actually, something like this seems what is argued in Sidelle 2001), I am curious about whether it is also necessary.
Finally, there are genuine metaphysical disputes, where none of the former applies (and I side with him against Karen that composition provides a nice example.)
Sider illustrates this with the case of ‘person.’ Let ‘person*’ be the entities individuated by psychological conditions, and ‘person#’ be the entities individuated by the bodily conditions. Every (relevant) disputant would agree that a description in terms of persons* and persons# is complete with respect to which (relevant) facts there are. The remaining issue is how there are to be described in terms of the older ‘person.’ If there is a semantic fact of the matter, then the dispute is genuine, though semantic. Otherwise, the dispute turns out not to be genuine after all, and all subsequent discussion should, I guess, be dismissed. Sider thinks the latter is the case, I tend to think that the former seems more plausible (see discussion in section 4), but this is another matter.