I don't think the argument depends on simulations being able to infinitely and recursively create sub-simulations. You only need one layer of simulations to mak

I don't think the argument depends on simulations being able to infinitely and recursively create sub-simulations. You only need one layer of simulations to make the odds work.