An argument against unprovable gods

This argument takes several premises, any one of which you may disagree with.

1. Science cannot possibly support the existence or non-existence of God.
2. Science can support any hypothesis over another if they make different predictions.
3. Occam's Razor: Given two hypotheses with identical predictions, we should always use the one that's easier to use.
4. The hypothesis that God doesn't exist is easier to use than the hypothesis that God does exist.
   
5. We should not believe in a hypothesis that we never should use.
   

Conclusion: We should not believe that God exists. (I think the reasoning is clear.)

Now, this is a deductive argument, meaning that it absolutely must follow from its premises. But such an argument is only as strong as its premises, and these are some questionable premises. Below, I discuss them one by one.

1. This premise is not my claim, but something I hear other people claim. I actually disagree with it. Note that here, "science" is used in the very wide sense of any process that can be used to choose between hypotheses.
2. If two hypotheses make different predictions, we can devise an experiment that will allow us to observe one prediction or the other. But in actuality, this premise is false. Perhaps the only possible experiment is impractical. For example, miracles are difficult or impossible to test in practice, since they are by definition non-reproducible events.
3. This is a very specific interpretation of Occam's Razor, as explained and justified in a previous post.
4. Let's just say that when I walk around, I don't even bother considering what effect an omnipresent but inactive god would have on my daily activities. If I did consider it, then I would simply end up reaching the same conclusions with more effort. If you're reaching different conclusions, you've either contradicted the first premise, or have demonstrated how confusing it can be to use the hypothesis that god exists.
   
5. I suspect this premise will get the most disagreement. Just because one hypothesis is simpler doesn't mean it's true, right? But the whole idea is that it doesn't matter whether it's true. A world in which it is true is indistinguishable from a world in which it is false. You can therefore believe what you want. Believing in an unusable hypothesis is a poor choice. Therefore, it is best to be agnostic or atheist with respect to this unprovable god.
   

I follow the same line of thought whenever I hear about "questions that science can't answer".

So, any disagreements?