It's actually not. See Gerard 't Hooft's "The Cellular Automaton Interpretation of Quantum Mechanics" .
Preferred reference frames aren't necessary, you just have to give up the assumption that experimenters are actually free to configure their experiments, aka superdeterminism.
The article refers to this derisively as a "cosmic conspiracy", which seems to be a common "attack" against this sort of interpretation, but it frankly has no teeth. The "cosmic conspiracy" charge is really just restating the fact that any particles with a common history are entangled, even particles in our brain, and since we all share a common history owing to the Big Bang, experimenters are not free to configure their experiments, but their decision is actually predetermined by this shared history.
Scientists have also gone on to claim that this sort of interpretation would make science impossible, but that's simply not true, any more than it's impossible for a deterministic algorithm like hill climbing to fully search a state space.
Suppose we are entangled in some state:
|we choose to measure observable A> @ |x> + |we choose B> @ |y>
Why would this determine which choice we make?
Yes, I said as much. That still doesn't entail that science is impossible though.
> Of course, [superdeterminism is] untestable
As is every other interpretation of QM.