This has been my favourite explanation since I read Stross' book, Accelerando. The common argument I get against it is along the lines of "humans being too curious to abandon the real universe".
Of course simulation in this case doesn't answer the problem of Von Neuman probes and the like, as its still perfectly reasonable to expect a simulated race to have some physical presence and desire to explore the universe, however since we recognize the risks of these kinds of things so long before we even have the capability of launching our own VN probes, perhaps the vanishingly unlikely scenario is an advanced civilization actually doing it themselves?
Of course simulation in this case doesn't answer the problem of Von Neuman probes and the like, as its still perfectly reasonable to expect a simulated race to have some physical presence and desire to explore the universe, however since we recognize the risks of these kinds of things so long before we even have the capability of launching our own VN probes, perhaps the vanishingly unlikely scenario is an advanced civilization actually doing it themselves?