Falsifying String Theory With Statistics

As many of you know the string landscape gives on the order of 10^500 solutions for possible universes predicted by string theory. One way to look at this is string theory seems to predict just about any universe and so has no predictive power. Furthermore, with this line of thinking you usually conclude that string theory is not falsifiable.

Another way to look at is is perhaps string theory predicts a multiverse with 10^500 different classes of universes. Some people, who take this view, go on to use the term "anthropic landscape" since immediately this way of thinking opens up the possibility that there are 10^500 universes but only a small subset support life and so our existence is natural since the theory alone with no fine tuning predicts intelligent life should naturally arise in some universes. (You say the odds of having such perfect constants that support life is highly improbable by chance alone, but when you have 10^500 such universes become highly probable.)

So, with this as a backdrop, enter Bousso and Leichenauer. They decided that you can now falsify string theory by showing that string theory favors universes that support life with different physical constants than ours. If string theory says: "Yes, some universes support life, but those that do should typically have physical constants different than those you see" than this in some sense would falsify string theory.

They found of all their universes with observers, the constants of our universe differ from the most probably constants by 2-3 sigma depending on the constant. They view this as a good sign for string theory.

I'm not at all saying "go endorse the string landscape." It's just an interesting way to try to falsify string theory.