Let me see if I can formalize. This might not be quite what you had in mind, but I think it will be similar:

For clarity we can reduce the possible worlds to two, either there are many many more Boltzman brains than human brains (H1) or there are few if any Boltzman brains (H2).

In H2 aprox. everyone who learns of the Boltzman brain hypothesis (and the evidence in favor) is not a Boltzman brain.

In H1 very very few Boltzman brains will learn of the Boltzman brain hypothesis (and the evidence in favor). A significantly larger percentage of the non-Boltzman brains capable of conceiving the hypothesis will learn of it (and the evidence in favor).

So independent evidence of H1 means (1) H1 is more likely to whatever degree that evidence dictates, (2) if H1 you are more likely than most brains to be non-Boltzman, (3) by the self-indication assumption H2 is more likely because in that world most or all brains are non-Boltzman.

The inference from (2) to (3) seems problematic to me. I'm not sure.

Questions:

1. How the hell do we evaluate the evidence since any evidence of H1 is also evidence of H2 (if we like the SIA).
2. What the hell is the proper reference class?
3. If new evidence came in against H1 would we have to say were more likely to be Boltzman brains?