The laws of physics are governed by equations which output only a probability distribution P(x) for various quantities (positions of particles, the values of certain fundamental constants).
(Note: I really don't understand any of this stuff well, so if someone (Dani?) with better knowledge can correct me or clarify, I'd love to see it.)
In a single-universe theory (SUT), one postulates that a particular x is selected, at random, relative to P(x) dx. In this theory, it is highly unlikely that the universe can support life (or at least life like us). In a multiverse theory (MUT), one postulates that all x are selected in parallel. In this case, our particular x will be chosen; the fact that we live in a good universe is not luck, merely natural selection. So the probability we exist is close to 1.
Both these theories are more or less equally "simple" (think Kolmogorov complexity), so let us assume they each "occur" with the same or similar "probability" (i.e., p(SUT)=p(MUT)=1/2). Then we use Bayes rule:
p(SUT | we exist) = p(we exist | SUT) p(SUT) / p(we exist) = (really freakin low) (1/2) / (1/2+really freakin low) = (unlikely)
p(MUT | we exist) = p(we exist | MUT) p(MUT) / p(we exist) = (almost 1) (1/2) / (1/2 + really freakin low) = (highly likely)
So multiverse theory is more "probable" than single universe theory. This remains true even if you wildly alter the complexity of competing theories, e.g. p(SUT)/p(MUT) = 1e6.
It's true, we can never falsify SUT or MUT relative to each other. But we can make educated guesses as to which one is more "probable". And yes, this is pretty sketchy, hence my use of scare quotes everywhere. But it's the best we can do right now.
Note: I computed p(we exist) = p(MUT)p(we exist | MUT) + p(SUT)p(we exist) = (1/2) 1 + (1/2)(really freakin low).
[edit: just curious why I'm being downmodded. Not begging for upmods, I'm honestly curious. Did I make a mistake here?]
I did a physics degree. I studied atomic physics and quantum mechanics. We briefly brushed on the multi/single universe, but largely that was not examined in too much detail in the physics cursus. People who were doing Physics & Philosophy, however, got plenty of detail in their course. My point being: this is philosophy, not science.
[edit: my turn to wonder why I am being downmodded, lol]
The question is 'absent plausible experiments, how do we choose between single and multiple universes?' And that is done using occams razor; the 'simpler' or 'more probable' theory is chosen, for a somewhat (but not completely) arbitrary definition of 'simpler'.
Occams razor is a philosophical point, as is (to some extent) your definition of 'simpler'. I thought I was careful to highlight those points.
I was merely trying to explain the reasoning multiverse theory, but in a less vague manner than the article did (I trust most of HN to know Bayes rule).
Incidentally, this thread is bringing out behavior which is weird for HN. Someone downmodded your comment, also for reasons which are completely inscrutable to me.
For the record, I haven't read the article in question closely enough to know if what's being proposed can even be falsified in principle.