Just to nit-pick, the simulation argument doesn't rely on simulations-in-simulations, let alone an "infinite regress". If we consider the proposed form of simu

Just to nit-pick, the simulation argument doesn't rely on simulations-in-simulations, let alone an "infinite regress". If we consider the proposed form of simulation to be feasible/reasonable/etc. then the existence of one simulation would give us 50/50 odds of being inside it. With two simulations running, we're more likely to be simulated than not. The simulation argument claims that, if such simulations are possible at all, then there will be very many of them. In which case, the improbability of being in a particular simulation is more than compensated by the number of simulations.