It is generally accepted that there is no experimental evidence that could determine between macroscopic decoherence and the collapse postulate. After all, figuring out whether worlds that we cannot possibly interact with exist is difficult. There’s definitely some ways to determine that many-worlds is correct- Eliezer wrote half a sequence on why Occam’s Razor suggests many-worlds- but it would be nice to have some evidence, some way to observe the way the world is and determine which interpretation is right.
The answer lies in a certain popular argument by creationists. They look at all of the many factors necessary for life to exist, and point out that the chances of these factors occurring randomly are very low. Surely there was some sort of process that caused these factors to be set just right… an intelligent process.
Now, the fact that we never interact with a god means that our prior probability of creationism would be low. Let’s say (these numbers are nowhere near exactly correct, but they illustrate the point) our prior probability of god is 1/1000, and the chance of life occurring by chance is 1/1,000,000,000.
Suppose that the collapse postulate was right, and there is only one world. In that case, the chance of us existing by chance is lower than the chance of god existing, and so posterior probability of god would be 1,000,000 to 1.
But if many-worlds is correct, then out of all the possible worlds, a few would have randomly generated life. Since all possible worlds have the same laws of physics, there wouldn’t be god in any of them unless there was god in all of them. Since it would be guaranteed, based on many-worlds, that life would exist without god, then the probability of god existing would remain the same (1/1000.)
So if many-worlds is correct, the probability of creationism is incredibly low, while if the collapse postulate is correct, the probability of creationism is incredibly high. We have observed none of the evidence that we would expect to see if creationism is correct, so Bayes’s Theorem says many-worlds is almost certainly true.