As has already been said, the only honest answer anybody can give at this point is that we just don’t know.
But that doesn’t mean that there isn’t an abundance of speculation, and some tantalizing hints from existing theories: for instance, it’s impossible to have a truly empty space – sure, you can clear out all matter and stuff, but you’d be left with all sorts of quantum fields and virtual particles; and if you somehow cleared that out, they’d spontaneously reform (out of ‘nothing’), because their presence is energetically favourable to their absence. (See here for a little more info on this.) True, this still presupposes the existence of some background space, and the validity of the laws of quantum field theory, but I think it’s an illuminating analogy.
Similarly, consider the creation of virtual particle pairs – ordinarily, they must always recombine, and ‘return’ the energy they borrowed; but, under certain conditions, such a particle can escape and become ‘real’ (though usually at a price – in Hawking radiation, for instance, an escaping particle has to ‘steal’ its energy from the mass of a black hole).
So, in an analogical way, it might be that nothing is just an unstable state of being, which ‘decays’ into something; however, there are some bootstrapping issues with this view – the rule that ‘something decays into nothing’ (and the physics to substantiate it) would seem to have to precede the existence of something, in order for the nothing to know that it must decay; but where, then, does that rule come from? There seems to be a curious teleology at work here: the nothing must decay into something in precisely such a way as to create a world in which nothing must decay into something!
But actually, there is a strikingly similar teleology at the heart of even classical mechanics, embedded in something called the principle of least action. According to this principle, the evolution of any given system follows a trajectory such that a certain quantity, called the action, is minimized (or more accurately, made stationary). What, exactly, this means isn’t very important right now; wiki does an alright job at explaining it if anybody’s interested. The thing is – how does the system ‘know’ to evolve in precisely such a way? It would almost seem that points in the future on this trajectory exert an influence on points in their past, ensuring that the ‘correct’ trajectory is traced out! This runs counter to all our accustomed notions of time and causality.
The answer to this conundrum comes, once again, out of the mysteries of the quantum world. Here, a system, in some sense, doesn’t have to decide right away which trajectory to follow – it gets to evolve, in parallel, along all possible trajectories (this notion is what ultimately led to Feynman’s path integral formulation of quantum mechanics). It is only when probed by some interaction that it has to make up its mind, so to speak – and this it does with overwhelming likelihood (for any remotely macroscopic system, practical certainty) in such a way as to satisfy the principle of least action, neatly eliminating the curious teleology.
Of course, this doesn’t nearly begin to solve the question, either. In a genuine ‘nothingness’, where should (some analogue of) the principle of least action come from? Why would it have to hold?
Well, I think there is some way to make this all appear a bit more plausible. Let’s consider a different way of looking at the evolution of the universe in time, one that, in a way, gets rid of time (with its always troublesome questions of what was ‘before’ etc.). Let’s think of the universe as a system of N particles; each of those particles has three spatial coordinates, so we can represent this universe in a 3N dimensional space, called the configuration space. Each point in this space represents a configuration of matter in the universe, and a trajectory through this space traces out the time-evolution of the universe. We can apply a special form of the action principle to this space, known as the Jacobi principle, which does not depend on time, but only on the 3N generalized coordinates, and in such a way obtain a unique history of the universe dependent only on its endpoints (this is taken from Julian Barbours ideas which he presents here (PDF)).
Now, generally, there’s nothing that says that the number of particles is to stay constant over time – for instance, if you pull hard enough on a meson, thanks to colour confinement, you’ll end up with two. Thus, it seems that the trajectory through the configuration space is only ever confined to a 3N dimensional subspace of a much larger (potentially infinite) space at a given point, but can evolve beyond that.
But then, nothing has a configuration space, too – a zero-dimensional one, because N is zero. However, now we have nothing that stops us from having our trajectory start in the zero dimensional subspace of the configuration space, and trace out a history according to Jacobi’s principle that corresponds to a universe in which matter is seemingly created out of nothing!
As a word of caution, though, this isn’t meant to be anything but an analogy for how it might be that something can seemingly come into existence; whether it works in detail or not (it probably won’t) isn’t all that important, this just goes to show that it isn’t inherently impossible to think about the concepts involved in some not totally nonsensical way. It’s just a bit of out-there speculation, is all.
Of course, there are other lines of thinking, as well – I recently read a proposal by Laura Mersini-Houghton that essentially amounts to treating the so-called ‘string landscape’ as ontologically real and eternal and having the wave function of the universe propagate across it in a sort of direction-less time, ‘settling’ into one of the roughly 10[sup]500[/sup] possible string vacua and thus giving birth to our universe; interestingly, this produces interactions with neighbouring universes due to quantum mechanical entanglement, which supposedly account for the recently-observed ‘dark flow’ and WMAP cold spot. Here’s a thread about it I killed.
As one last note in an already overlong post, I don’t believe it’s the case that you have to choose between the two options of ‘eternal existence’ and ‘creation out of nothing’ – there are other alternatives possible; I’ve seen proposals involving circular time, both on the large and the small scale, with our universe being kind of a linear spin-off of the circular region in the latter case, counterpointing arrows of time, where the big bang kind of serves as the mid point between two universes evolving in opposite directions in time, and many more.
So, here’s to hoping this won’t kill the thread, I’ve kinda had a bit of a bad track record lately…