David Lewis's Counterfactuals is an interesting work, because it has many interesting insights but one can also see in it the beginning of a number of pathologies that have become common in analytic use of possible worlds. Lewis glosses 'possible worlds' as "'ways things could have been'" (p. 84) but also wants to say that our actual world is exactly one of the possible worlds:
Our actual world is only one world among others. We call it alone actual not because it differs in kind from all the rest but because it is the world we inhabit. The inhabitants of other worlds may truly call heir own worlds actual, if they mean by 'actual' what we do; for the meaning we give to 'actual' is such that it refers at any world i to that world i itself. (pp. 85-86)
This is (unlike what you often find) a correct interpretation of how the 'actuality' operator works in cases in which it is added to possible world semantics; it just established a privileged world, for whatever reason one wishes. This is, I think, a good reason for thinking that the 'actuality operator' no more captures the actuality of the actual world than the 'existential quantifier' captures the existence of an existing thing; but set this aside. However, there is a tension between thinking in this way and thinking of possible worlds as 'ways things could have been'. We see this in an inadvertent slip:
Among my common opinions that philosophy must respect (if it is to deserve credence) are not only my naive belief in tables and chairs, but also my naive belief that these tables and chairs might have been otherwise arranged. Realism about possible worlds is an attempt, the only successful attempt I know of, to systematize these preexisting modal opinions. (p. 88)
These tables and chairs might have been otherwise arranged. Which ones? The ones in the actual world, which Lewis insists is just one possible world, and therefore cannot possibly be otherwise than it is.
This is, of course, the reason why Lewis has his famous difficulties with transworld identity. Considering the case of Ripov, who does not bribe a judge but could have done so, he sets up a dilemma. Either there has to be transworld identity -- one and the same Ripov exists in multiple worlds -- or a counterpart relation -- the Ripov of this world is similar although different to the Ripov of another world. Lewis, while recognizing that there is some inconvenience to it, chooses the second:
The best thing to do, I think, is to escape the problems of transworld identity by insisting that there is nothing that inhabits more than one world....Things that do inhabit worlds -- people, flames, buildings, puddles, concrete particulars generally -- inhabit one world each, no more. Our Ripov is a man of our world, who does not reappear elsewhere. Other worlds may have Ripovs of their own, but none of these is our Ripov. Rather, they are counterparts of our Ripov....What our Ripov cannot do in person at other worlds, not being present there to do it, he may do vicariously through his counterparts. (p. 39)
Thus, this Ripov cannot in fact do otherwise than he does; the different possible worlds do not capture the ways this Ripov could have been. Why then are they relevant? They're just different worlds, ghost worlds from our perspective, but different worlds. The root problem here is thinking of possible worlds as worlds at all. They are not ghostly other worlds. How could they be? If they were ghostly other worlds, how would we know anything about them? Why would they be relevant to what happens in our world, any more than having a doppelganger across the ocean says anything about your own modalities? Yet it takes no extensive reading to see that this treatment of possible worlds, that other possible worlds are phantom worlds separate from our own, completely distinct from our own, and that there are ghost-people and ghost-things existing in these ghost-worlds, about which we have some sort of magical clairvoyant knowledge, is found all throughout analytic discussions of modality.
The peculiarity of all of this becomes more clear when we consider why Lewis rejects transworld identity. He holds that the problem is that the identity is just mysterious, "an irreducible fact, not to be explained in terms of anything else" (p. 39) or else just has to reduce to the kind of resemblance that fits better with the counterpart relation. But there is no mystery. The Ripov in another possible world is the Ripov in the actual world, because he is a way this actual Ripov could be. The whole point, the whole point, of talking about possible worlds as 'ways things could be' is to give the underlying logical structure of what we say about the actual Ripov. We did not discover the ghost worlds like galaxies through a metaphysical telescope; we experienced the actual world and recognized that things in it cannot be properly accounted for or placed in a coherent narrative unless we recognize that the actual things themselves could be otherwise -- that this Ripov who doesn't bribe the judge could have. If you think of possible worlds as worlds at all, they have to be treated as the actual world itself, considered in a very selective way. These tables and chairs might have been different; this Ripov might have been different. Ripov has transworld identity because Ripovs in other possible worlds are just descriptions of the possibilities of Ripov.
Of course, as I've noted before, nothing about the actual formal system requires that we think of possible worlds as worlds at all; they are logical entities mapped to consistent sets of truth-valued logical propositions, which we can interpret in endless numbers of ways. Even interpreted as 'ways the actual world could have been', however, nothing requires that they be treated as worlds; as ways the actual world could have been, it makes more sense to treat the actual world as the world, and the possible worlds as descriptively capturing 'slices' of that very world. They aren't ghost worlds; rather, they are 'beings of reason', and we use them to describe the actual world in which we live.
****
David K. Lewis, Counterfactuals, Harvard UP (Cambridge, MA: 1973).
