Is the future in some sense produced or generated from the present? Our intuitive folk metaphysics seems to assume so. The present moment is taken to be metaphysically privileged -- until the laws of nature do their work, generating the 'next' moment out of the raw materials of the present one. This generative process constitutes the passage of time, as the past is literally transformed (in the deepest fabric of its being) into the future. According to this 'presentism', the old times are not simply passed, but wholly replaced.
It's a difficult subject to make sense of. We're used to talking about objects within time, and it's not so clear that our familiar concepts (process, change, replacement), which compare an object between times, can be extended to apply to time itself -- not without regressive appeal to some 'meta-time' during which the manipulation of time-as-an-object occurs. (More here.)
These concerns lead us towards 'Eternalism', or the view that all moments are ontologically on a par. Time is understood to be simply another dimension, not so different from space, and 'now' is an indexical with no more metaphysical import than 'here'.
It seems that this forces us to give up our intuitive beliefs about metaphysical production. If the future already exists, it doesn't need to be generated out of the present. Laws of nature are stripped of their productive power, and instead serve the merely passive purpose of descriptive generalization.
Consider the 'Humean mosaic': the static, 4-d spacetime "loaf" that comprises the entire expanse and (future) history of the universe. The distribution of properties is taken to be fundamental and inexplicable. The way of things is just a brute fact. It just so happens that there are plenty of regularities -- "constant conjunctions" -- to be found in the mosaic. Dropping an object is followed by its falling to the floor. There's not really any deep reason for this (it's just the way the mosaic happens to be drawn), but we can come up with physical "laws" that describe these regularities.
So, what are we to make of all this? Can the luscious garden of common-sense metaphysics be saved? Or should we embrace the desert landscape of the Humean mosaic? I guess David Lewis' project was to show how, with the help of some fancy analytical footwork, we can still assert the truths of common-sense within the desert. Present events still "cause" future ones, and all that. It's just that causation isn't what we thought it was. (Rather than a fundamental notion of generative powers, where the cause brings about its effect, we appeal to a sterile analysis in terms of counterfactual conditionals, which are in turn reduced to facts about other - similarly 'deserted' - Humean mosaics.)
I'm awfully suspicious of that kind of philosophy, as explained in my old post: The Limits of Truth Conditions. I guess I'm assuming some kind of semantic transparency here: we have some idea of what it would take to make our claims true, and the desert just ain't it. I mean, nobody ever read Lewis' On the Plurality of Worlds and thought, "Exactly, that's what I meant when I said that Humphrey could have won the election: just that somewhere out there, in another universe, there's a guy a lot like Humphrey who does win a similar election!"
So I have some secondary questions: Should we be satisfied by crazily counterintuitive analyses, so long as they yield the (truth-conditionally) right results? Why allow that common-sense intuitions are a guide to what's true, but not to what those truths consist in (i.e. what's real)? On the other hand, is it a legitimate objection to Lewis to simply insist, "But that's not what I mean by that term!" -- is meaning so introspectable? How much leeway do we have when assigning truthmakers to a class of claims?