Sean Carroll on falsification

A very interesting essay, worth reading in its entirety if you have the time. Some selected highlights for those that don't.

In its strongest from, the objection argues that the very idea of an unobservable multiverse shouldn't count as science at all, often appealing to Karl Popper's dictum that a theory should be falsifiable to be considered scientific. At the same time, proponents of the multiverse (and its partner in crime, the anthropic principle) will sometimes argue that while multiverse cosmologies are definitely part of science, they represent a new kind of science, "a deep change of paradigm that revolutionizes our understanding of nature and opens new fields of possible scientific thought".

In this essay I will stake out a judicious middle position... The point is not that we are changing the nature of science by allowing unfalsifiable hypotheses into our purview. The point is that "falsifiability" was never the way that scientifictheories were judged (although scientists have often talked as if it were)... The best outcome of current controversies over the multiverse and related ideas (other than the hopeful prospect of finding the correct description of nature) is if working scientists are nudged toward accepting a somewhat more nuanced and accurate picture of scientific practice.

Science proceeds via an ongoing dialogue between theory and experiment, searching for the best possible understanding, rather than cleanly lopping off falsified theories one by one. (Popper himself thought Marxism had started out scientific, but had become unfalsifiable over time as its predictions failed to come true.) While philosophers of science have long since moved past falsifiability as a simple solution to the demarcation problem, many scientists have seized on it with gusto, going so far as to argue that falsifiability is manifestly a central part of the definition of science.

The multiverse, therefore, is a case of science as usual: we evaluate it on the basis of how likely it is to be true, given what we know on the basis of what we actually have observed. But it is not only examples of literal new data that can cause our credences in a theory to change. The multiverse hypothesis reminds us of how better understanding, as well as actual experimental or observational input, can serve as "data" for the purposes of Bayesian inference... A correct accounting for the multitude of influences that shape our credences concerning scientific hypotheses is in no sense a repudiation of empiricism; it is simply an acknowledgment of the way it works in the real world.

The best reason for classifying the multiverse as a straightforwardly scientific theory is that we don't have any choice. This is the case for any hypothesis that satisfies two criteria: 1) It might be true. 2)Whether or not it is true affects how we understand what we observe.

None of which is to say that there aren't special challenges posed by the multiverse. At a technical level, we have the measure problem: given an infinite multiverse, how do we calculate the relative probabilities of different local conditions? Skeptics will sometimes say that since everything happens somewhere in the multiverse, it is impossible to make even probabilistic predictions. Neither of these two clauses is necessarily correct; even if a multiverse is infinitely big, it does not follow that everything happens, and even if everything happens, it does not follow that there are no rules for the relative frequencies with which things happen... There still remains the question, even if there is a correct measure on the multiverse, how will we ever know? It seems hard to imagine doing experiments to provide an answer. That, in a nutshell, is the biggest challenge posed by the prospect of the multiverse. It is not that the theory is unscientific, or that it is impossible to evaluate it. It's that evaluating it is hard.

Well, I remain a skeptic. I don't see how you get a meaningful explanation of anything if you throw an infinity in there - to me, that's a sign of a theoretical failure. Sure, the world might be like that, but that would basically reduce it to being fundamentally illogical. I can't conceive of how you can have meaningful probabilities in an infinite reality. Maybe you can, but it feels an awful lot like cheating to me.

IMHO, string theory, inflation, the multiverse and all that jazz are not straightforward, "business as usual" scientific theories. They're a far cry from being pseudoscientific bunk, but neither are they comparable to relativity or Maxwell's eqauations. They hardly represent some wonderful new advanced sort of science, rather, they are more in the fuzzy grey area between genuine pseudoscientific woo and hard physical reality. See, while I agree wholeheartedly that "falsification" can't be the whole definition of a scientific theory, I do hold that the prospect of falsifiability is always better. That doesn't lead to anything so binary as "non-falsifiable is bad, falsifiable is good", let alone scientific/unscientific. But it does suggest that there is something at least different about unfalsifiable ideas like the multiverse.

I maintain that a scientific idea should be at least testable, if not falsifiable. You have to be able to evaluate its success with respect to observational evidence and other competing theories. The multiverse might satisfy that criteria, but currently string theory does not. That doesn't preclude them from ever becoming testable, however. Perhaps we should think of them as proto-science.

See also : http://astrorhysy.blogspot.cz/2017/05/i-told-you-he-was-tricksy.html
https://arxiv.org/abs/1801.05016