Lets imagine that the mathematics of our ToE suggests the existence of numerous empirically undetectable universes. This same mathematics is used for all sorts of practical applications and accurate predictive results in our own universe which establish it truly as a theory of everything.
Fundamental to this mathematics is the existence of universes that have no practical effect on our universe.
Sounds like this hypothetical ToE has a lot going for it. I would doubt then if the theory would include something for which there was no effect in our universe. If such a ToE showed that the multiverse was probable then it would also predict a testable effect. We are talking a strong predictive theory here not a hypothesis. I cannot conceive that it would include in its mathematical model something which had, by definition, no effect to model.
But let's play the game.
If the theory predicted something that had not, and could not have, any discernable effect on our universe then the math is predicting something that cannot be said to exist. It wouldn't matter whether this thing was real or not. For us, it has no effect, it means nothing, and therefore might as well not exist.
Now if you're thinking that maybe based on the knowledge of this non-effective thing we develop further hypotheses that may have some practical discernable effect or use to us, then this would belie the premise that the hidden non-discernable thing was actually hidden and non-discernable. It could be tested by way of these new hypotheses because of their effect in our universe.
But to answer the question specifically, if by some strange twist of math the ToE modeled something for which there could not possibly be a model then there is a hole in this beautiful and consistent theory somewhere.
Edited by AZPaul3, : The usual typo suspects.
™ Version 4.2