Sean Carroll in Preposterous Universe:
Inflationary cosmology is the clever idea that the early universe underwent a brief period of accelerated expansion at an enormously high energy density, before that energy converted in a flash into ordinary hot matter and radiation. Inflation helps explain the observed large-scale smoothness of the universe, as well as the absence of unwanted relics such as magnetic monopoles. Most excitingly, quantum fluctuations during the inflationary period can be amplified to density perturbations that seed the formation of galaxies and large-scale structure in the universe.
That’s the good news. The bad news — or anyway, an additional piece of news, which you may choose to interpret as good or bad, depending on how you feel about these things — is that inflation doesn’t stop there. In a wide variety of models (not necessarily all), the inflationary energy converts into matter and radiation in some places, but in other places inflation just keeps going, and quantum fluctuations ensure that this process will keep happening forever — “eternal inflation.” (At some point I was a bit skeptical of the conventional story of eternal inflation via quantum fluctuations, but recently Kim Boddy and Jason Pollack and I verified to our satisfaction that you can do the decoherence calculations carefully and it all works out okay.) That’s the kind of thing, as we all know, that can lead to a multiverse.
Here’s where things become very tense and emotional. To some folks, the multiverse is great. It implies that there are very different physical conditions in different parts of the cosmos, which means that the anthropic principle kicks in, which might in turn imply a simple explanation for otherwise puzzling features of our observed universe, such as the value of the cosmological constant. To others, it’s a disaster. The existence of infinitely many regions of spacetime, each with potentially different local conditions, suggests that anything is possible, and therefore that inflation doesn’t make any predictions, and hence that it isn’t really science.