Originally posted by Eh
Eternal inflation is very similar to the old steady state universe, even though proponents of the theory would like to downplay the similarities. Each "bubble universe" is created from a pre-existing space-time, but the universe as a whole may be eternal, as you've said. But the problem is, the pre-existing space itself is expanding. For a universe finite in size, logically it follows that the volume of space would be smaller and smaller the further back in time you look. Eventually at some time in the past, the entire universe (or multiverse, if you will) shrinks down to a point of infinite density. So even an eternal inflation universe would have a beginning. Andrei Linde admits even the self reproducing universe may start off with a primordial singularity.
But with an infinite volume of space, there may indeed be no beginning at all. However, philosophically there is something very unsettling about an infinite volume of space that has been expanding for an infinite amount of time. Maybe it's because there is comfort in the finite, or maybe it's just because infinity is such an ugly concept. Should analysis of the WMAP survey find conlusive evidence that the universe is actually finite, then this model will be in the toilet. That is, unless someone develops a model of inflation based on extra dimensional space that has a minimum of 4 spatial dimensions.
But ignoring WMAP for a moment, try to compare an infinite chaotic inflation universe with the old steady state models. Hoyle invisioned an infinite universe that had been expanding forever, but with matter continously being created from the vacuum to fill in the great voids. Later models dropped the perfect cosmological principle (that idea that the universe looks the same everywhere and at all times) and included the ideas of mini big bangs arising from the vacuum. Though the math and mechanism for these later models were obviously different than inflation, the basic concept is very much alike. They both involved infinite space, and both involved an infinite regress.
At any rate, it seems the issue again comes down to the concept of infinity. Other models such as the cyclic universe, based on M theory also includes a universe that is infinite in space and time, so it seems to be an unavoidable topic these days.