As philosophers, we are often in the habit of making claims like the following: "it is impossible for a brick to just pop into existence uncaused out of nothing." Yet, possible worlds semantics throws a monkey wrench into this proposition. For, what if there is a possible world in which things, including bricks, are capable of popping into existence uncaused out of nothing? Most of us do not believe this can happen in the actual world, but it appears to beg the question if we map this principle onto other possible worlds.
This brings up the distinction between metaphysical possibility and logical possibility. It is impossible for me to fly by flapping my arms, but in what sense is this really impossible? Specifically, what is implied by this observation is that it is metaphysically impossible for me to fly by flapping my arms - that is, it is impossible in the actual world. However, one may very well postulate some possible world in which I can fly by flapping my arms, and perhaps this is because there are different natural laws governing that world.
However, the ability to postulate something (or the fact that one may conceive of some possible world in which X is the case), I believe, does not always establish that it really is possible, even in the strictly logical sense. This is especially pertinent whenever we are discussing the existence of a logically necessary being, or a being that exists in all possible worlds. One may very well conceive of a possible world with no necessary being, and without any contradiction to be found, it is said that this truly is a possible world. And, if a necessary being does not exist in any one possible world, it follows that it does not exist in any possible world at all, the actual world included. After all, if a necessary being exists in only some, but not all, possible worlds, it is not necessary to begin with, but contingent.*
Let's turn the tables now, though. It is just as easy for me to conceive of a possible world in which a necessary being does exist, and neither does this appear to be contradictory. Given S5, it follows that a necessary being actually exists. Yet, the opponent of this argument will not accept that a necessary being actually exists. This implies that possibility is not determined by mere conceivability, since the same standard can be used to establish mutually exclusive propositions. Think of it this way:
1. If X is conceivable, then X is possible.
2. It is conceivable that a necessary being does not exist in some possible world.
3. Hence, there is a possible world in which no necessary being exists.
4. Therefore, a necessary being does not exist. (This only follows once "necessity" is understood as logical necessity)
5. It is conceivable that a necessary being exists in some possible world.
6. Hence, there is a possible world in which a necessary being exists.
7. Therefore, a necessary being exists.
(4) and (7) contradict one another, which means that something must be wrong with our starting principle in (1).
*This objection can only apply to arguments that attempt to establish, say, God's logical necessity. They are impertinent to God's metaphysical necessity.