There have been a few threads of late on the multiverse concept in cosmology, and whether it can be viewed as a viable, albeit currently underconstrained cosmological theory that is leading us to demonstrably correct discoveries about our universe, or if it is essentially a fairly arbitrary metaphysical conviction that is masquerading as science. I'd like to advance the latter thesis, and central to my argument is the Popperian stance that if, as Feynman said, science should be a way to keep us from fooling ourselves, then we need theories that make "risky" predictions-- predictions that, were we to be skeptical of the theory, we would expect to fail. A theory that only makes predictions that no one can expect to fail, even if they discount the theory, is more like a technique for performing rationalizations than it is a technique for making predictions. To allow proponents to attack that thesis, I will offer an analogy. I hold that if my thesis is flawed, then its flaws should be identifiable with flaws in this analogy. Let's say you see five cards in front of you, and you are told the rules of a new game, called phyzbin, a game that you have never played before. You pick up the hand, and see you have 2, 4, 9, jack, queen. You know this is an extremely special hand in phyzbin, called a phyzlaw, because the rules of phyzbin select this hand to be the best possible-- there is a significant selection effect that has occured here, sometimes expressed that this hand is "finely tuned" in regard to the rules of phyzbin. We need to attribute a source to this selection, we cannot accept it was "just chance" because the odds are too low. But there are many ways to attribute that selection, including: 1) I was selected from a huge set of people playing the game (the people are analogous to the multiverse). There is something special about me, among all those people, such that I get the phyzlaw. Perhaps it is a requirement for me to be who I am, my existence is predicated on being the one who has such a hand. In anthropic thinking, we might say that only the phyzlaw is consistent with me being intelligent. Whatever is the mechanism, the key point is that I have been selected, and that resolves the "fine tuning problem." Instead of asking what is special about the hand, we ask what is special about me such that the hand appears more generic. The other hands must really exist, such that my hand can be selected from them all, because the selection applies after the hands are dealt, it is a selection on me, not on the hands. This seems like a very close analogy to multiverse thinking, and it is certainly one possible "explanation" of the fine tuning. But it is hardly the only possibility. 2a) The rules of phyzbin were selected after ony one hand was dealt, such that a phyzlaw would be that hand. Of course we know that any hand is incredibly unlikely if you deal from 52 cards, so the specialness of the hand only appears if we add some additional constraint that sorts all the hypothetical hands into "generic" and "special" classes. The rules of phyzbin produce that sorting, but if the rules are themselves selected, such that the hand is made to be special by those rules, then we again have no fine tuning problem. The question now shifts from what is special about the hand, to what is special about the rules such that the hand could be viewed as generic. I would argue that this view is analogous to a view of physics that says laws are not fundamental to the universe, they are what we infer in our efforts to understand the universe. Hence, the laws come from us, so it is perfectly natural that the laws share our own special attributes. If there is something about us that says we must have hands like 2-4-9-j-q, then the "rules of the game" that we infer will naturally make a phyzlaw seem like a generic hand, and "poof" goes the fine tuning problem. 2b) The deck contains only 2s, 4s, 9s, js, and qs, so we are simply making a wrong assumption about the possible hands when we conclude a phyzlaw is a special hand. This is similar to #2a, but distinguishes the rules of the game from the elements that make up the game, a distinction we can make if we like but might not need to. 3) The deal was rigged to give us a phyzlaw, perhaps because someone in power wanted us to win. This is a preferred choice of religious people. Now, as scientists we can probably dispense with #3, not because we know it's wrong, but because it's not in the scientific toolkit. And #2b is a kind of variation on the theme of #2a, so I see the fundamental distinction as being between #1 and #2a, and even then, only once we have established that the "fine tuning problem" really is an issue that science needs to tackle, on grounds that tackling it will help us make "risky" predictions. I would argue two points that undercut the idea that the multiverse is not just metaphysics: i) There is no scientific way to distinguish between #1 and #2, the distinction is about how we interpret science, and is therefore essentially metaphysical. ii) Even if we choose a mainstream interpretation that adjudicates for #1 over #2, it is still essentially an issue of pedagogy, because it does not help us use our perspective to make "risky" predictions that could satisfy a skeptic. What's more, being skeptical is a crucial part of science, because it is exactly that element that, as Feynman put it, helps us "avoid fooling ourselves, given that we are the easiest people to fool."