David Lewis, answered from the texts and cited to the page.
There are many other ways that a world could have been besides the way that this world is. Each such complete way is a possible world, and these worlds are concrete things — as real and as varied as our own, populated by their own people and stars and seas and quarks. They are spatiotemporally and causally isolated from us and from one another: nothing here is at any distance or in any direction from anything in another world, no event here causes any event there, and no observation can establish either way.1
Actuality is not a privileged metaphysical status that singles one world out; "actual" is indexical, picking out whichever world the speaker is at, just as "here" picks out whichever place the speaker occupies. Possibility for an object is not transworld identity (the same object existing across many worlds) but counterpart relation (objects in other worlds similar in relevant respects to this one), since the same object existing with different intrinsic properties in different worlds would replay the problem of temporary intrinsics at modal scale.234
The grounds for believing all of this are not direct (the worlds are by hypothesis causally isolated and so unobservable) but theoretical: the apparatus of concrete possibilia underwrites unified analyses of necessity, counterfactuals, propositions as sets of worlds, properties as functions from worlds to extensions, content, chance, supervenience, and much else, and the gain in unity-and-economy of total philosophical theory is sufficient warrant for the ontological commitment — on the same pattern of argument by which we accept set theory for the benefits it brings to mathematics.5
But if worlds are causally isolated, nothing outside a world ever makes a world; and nothing inside makes the whole of a world, for that would be an impossible kind of selfcausation. We make languages and concepts and descriptions and imaginary representations that apply to worlds. We make stipulations that select some worlds rather than others for our attention.On the Plurality of Worlds, pp. 11–12
On being part of versus existing according to, see section 4.1. 2.1 Everything is Actual? In this chapter of objections and replies, I begin with four objections meant to show that my modal realism, with the aid of uncontroversial auxiliary premises, leads to outright paradox. I reply by controverting the auxiliary premises.On the Plurality of Worlds, pp. 105–106
But insofar as he was trying to avoid all need a counterpart relation, as is suggested by his discussion of Catiline and the Great Pyramid, our reasons are not the same. If I thought it could be done, I would be glad to rescue the doctrine of propositional objects by means of a counterpart relation.Part 2: Philosophy of Mind, pp. 148–149
Our actual world is only one world among others. We call it alone actual not because it differs in irind from all the rest but because it is the world we inhabit. The inhabitants of other worlds may truly call their own * actual' worlds actual, if they mean by what we do; for the meaning we * actual' give to is such that it refers at any world / to that world i itself.Counterfactuals, p. 98
Or the physical constants might have had somewhat different values, incompatible with the emergence of life. Or there might have been altogether different laws of nature; and instead of electrons and quarks, there might have been alien particles, without charge or mass or spin but with alien physical properties that nothing in this world shares.On the Plurality of Worlds, p. 10
The most famous and most contested of my theses, and the one that earns the "incredulous stare" as standard reaction. The contemporary alternatives are the three families of ersatz modal realism — linguistic ersatzism (worlds as set-theoretic constructions out of sentences, as in Adams's world-stories), pictorial ersatzism (worlds as representations that resemble what they represent), and magical ersatzism (worlds as simple abstract objects bearing primitive selection or obtaining relations); each takes worlds to be abstract representations rather than concrete cosmoi. The actualist tradition (Plantinga's Nature of Necessity 1974, Stalnaker's Inquiry 1984, Robert Adams) holds that only the actual world is concrete and the others are abstract states-of-affairs or maximal-consistent propositions. Kripke's Naming and Necessity (1980) and the broader transworld-identity tradition rejects counterpart theory as failing to capture genuine de re modality. Sider, Nolan, and Hawthorne defend the systematic apparatus while diverging on individual doctrines.