As an example of the low-ontologist side in the constitution case, Bennett considers the Lewisian position that (let me over simplify here) even, if Statue and the Lump are identical, we can truly say that Statue would not survive being squashed into a ball while Lump would not by appealing to the different counterpart relations in which Lump/Statue stands with otherworldly things.
The heart of this strategy is to say that the relatively straightforward predicate 'being possibly squashed' in fact hides a multiplicity of more complex predicates that pack in some reference to the kind. (Lewis, of course, will invoke counterpart-theoretical properties like having a squashed counterpart under the lump-counterpart relation) Perhaps this require that the one-thinger [i.e. the one who takes the low-ontologist side in the constitution dispute] postulate a different complicated modal property for each object the multi-thinger [i.e. the one who takes the high-ontology side in the constitution dispute] countenances. Perhaps it just requires that she employ a different complicated modal predicate for each such object. That depends on the broader question about the viability of nominalism. What matters for my purposes is that the multi-thinger need not do either. (p.28)In other words, what the low-ontologists saves on the cost of her ontology comes at the price of her ideology. Now, I have no sympathy for Lewis' modal realism or his counterpart theory, but Bennett's interpretation of the Lewisian position does not seem to be particularly charitable to me. Let me put aside the issue of nominalism and that of conceptual vs. ontological simplicity and focus on Bennett's interpretation of the Lewisian use of the counterpart relation in this case.
As far as I can see, the Lewisian's reply to Bennett should be that he does not need the complex predicates or the corresponding properties. When saying that 'Lump would survive being squashed' is true and 'Statue would survive being squashed' is not even if 'Lump' and 'Statue' refer to one and only one thing, the Lewisian would not directly appeal to the fact that the same thing has two different modal properties but to the fact that in different contexts the same thing can have different counterparts because the different contexts make different respects of similarirty with otherworldly things relevant. So, for example, when talking of Lump/Statue as 'Lump', we are making the material is made of, its mass, etc. salient, while when talking of it as 'Statue', we are making also its shape and history salient. So, there are things that are counteraprts of Lump/Statue qua lump of clay that are not counterparts of it qua statue (things that resemble it in being made of clay and having a certain mass, etc. but not in having a certain shape etc.) and some of this things are temporal parts of things whose other temporal parts were counterparts of Lump/Statue qua statue but are no longer counterparts of it because they no longer bear the right sort of resemblance to Lump/Statue qua statue because they have been squashed. So, the Lewisian really only needs the property having been squashed and claim that some counterparts of Lump/Staute qua lump of clay have it while some counterparts of it qua statue do not have it. It is only in virute of its counterparts having or not having the property having been squashed that Lump/Statue has or has not (derivatively) the modal property of being possibly squashed. Of course, Bennett could claim that the counterpart theory already comes at too high an ideological cost (I would just say that it is false, but I won't argue for that here), but Lewis and the Lewisians would claim it's a cost worth paying because of the benefit that it brings with it and, in any case, the Lewisian does not seem to need the strange predicates Bennett wants to saddle them with. Am I being too charitable to the Lewisian position or unfair to Bennett's objection?