Personally I find the idea of a completely deterministic system too depressing to bear, so I choose to believe in the free/will randomness model. Until there are solutions to things like the three body problem (both classical and quantum versions) and we can explain things like quantum foam spontaneous particle creation with a closed set, I'm probably gonna stick with this one. I'm also happy with the idea that a non-closed set (i.e. the universe is either infinite or has in/out relation with something else) implies omniscience is impossible since temporal omniscience requires computability which requires a non-finite set smaller than the complexity of the simulator if it is to run at real time or greater (see Turing's work on emulation and theory of computation). By extension omnipotence is impossible since it requires omniscience to know what's out there to have control over.
I consider the Many Worlds Theorem a crock though. It seems like a cute idea when you're using it as a quantum concept applied to a classical framework. But when you start extending it to apply universally to a quantum system, it gets ugly. For example, people like to say "well every event makes new realities for the different outcomes". Ignoring the fact that this means conservation principles are not transcendent (i.e. conservation of matter and energy should imply conservation of reality-space at the next order), there's the problem of the number of outcomes in a quantum system being non-finite for each event or potential event. This last part's the killer, let me try to explain.
If you flip a coin, basic old style math says it's a 50/50 heads or tails chance for the result. The commonly used catch for this one is "coin lands on its edge". But this is still a classical system result. When we start expanding the system to incorporate all possibilities regardless of probability (like say, the coin gets destroyed by a passing cosmic ray burst), then it gets real bad. Because many worlds relies on this whole concept of "exhaustive execution", i.e. "Anything that can happen, does happen, in some reality, somewhere." But this implies that low probability events also occur, in fact they also must occur in infinite number, so we should see low probability events all the time. We're bound to run into them eventually as they're all over the place. Some people like to use this to explain events they consider strange. But the broader implications are disturbing.
For example, if all potential events are exhaustively executed, it means it's just sheer luck that whatever you'd like to consider your consciousness or awareness (or however you'd like to term it) hasn't fallen into one of the low-probability paths. Some of these lower probabilities may be pleasant, like winning the lottery. Others may not, like stabbing the person you love to death with a knife without provocation. The argument against this is usually "But that's too improbable to happen, I would never do it." But the thing is exhaustive infinity is mutually exclusive with free will. You will do it. Because you must do it. And in fact you already have done it, in an infinite number of realities that you just barely managed to avoid winding up in. So basically it's completely chance you avoid not only experiencing every possible horror you can imagine each second, but also being forced to commit unspeakable acts without reason. And this could change without warning at any instant.
Personally I find that unacceptable. I'd rather believe in a non-exhaustive expression system where low probabilities are culled, resulting in a still infinite set but with certain areas avoided (see fractals for an example of how you can have an infinity with holes in it).