Consider the usual derivation of Hawking radiation, using quantum field theory (QFT) in the spacetime of a collapsing star. At late times, long after the event horizon has formed, Hawking radiation is independent of the details of the collapse that formed the event horizon. This independence famously leads to the information loss paradox. However, it still depends on which QFT we use. For example, if we use a QFT that doesn't have any spin-$1$ particles in flat spacetime, then the Hawking radiation won't contain any spin-$1$ particles, either.
According to the landscape concept, string theory has lots of different "vacua" — lots of different possible low-energy effective theories. Since the particle content of Hawking radiation can be derived using the low-energy effective QFT, it must also depend on which point we choose in the string-theory landscape when Hawking radiation is derived using string theory.
Is this right? I'm no string theory expert, but I had the impression that black holes in string theory were excellent scramblers, taking whatever goes in and scrambling it beyond practical recognition (even if the information is still recoverable in principle). But the preceding argument seems to say that the scrambling abilities of a black hole are limited: they can't scramble different points in the landscape with each other. Is this right?
Or can black holes mix different points in the landscape, too? If so, then where are the flaw(s) in my reasoning?