For the Universe as a whole system, there is no environment and no history to set boundary conditions. This has led many cosmologists to speculate that the universe we perceive might be a tiny “bubble” in a much vaster plurality of universes. If such a “multiverse” were large enough, perhaps infinite or eternal, the problem of initial conditions would disappear. Enough trial runs of random “universes” would inevitably produce one like ours. Many scientists are not satisfied, however, to let the beginning go thus unexplained in traditional causal terms. Some question the extravagance of bringing in a metaphysically extravagant multiverse to contend with improbable initial conditions, which simply defers the issue of the origin. It would be ontologically more economical to question the standard model or the ways in which it is applied to cosmological observations. Astronomy is a highly inferential science. Nominally empirical measurements are actually highly theory-dependent. The estimations of improbability inspired by them may be spurious. Indeed, the very notion of a randomly generated universe is nonsensical if there is but one role of the cosmic die. Supposed implications of “fine-tuning” may reflect the state of science more than the state of the universe.

Multiverses come in several varieties. One concept has “universes,” such as ours, as particular regions in a meta-space. Given special relativity, these would most likely be causally isolated from each other, which makes such a scheme difficult to falsify. Another concept has successive universes in a meta-time. These would also be causally isolated if information is not preserved from one universe to the next. Other versions might combine features of such schemes—for example, Lee Smolin’s black hole cosmic selection theory. One motivation for all such schemes is to solve certain conundrums in cosmology, such as the appearance of fine-tuning of various parameters of theory required to predict a universe like ours hospitable to life.