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.
Of course as someone who writes simulations I can tell you that it is not so easy in our computational models. Indeed it would seem very difficult, though not fundamentally impossible for a high level model to correct macro behavior in all cases. But we cannot know the constraints of the universe which contains our simulation, which may be very different from ours.
The point is that unless there's some fatal logical flaw in the set of choices, some combination of the choices needs to be true, and whichever choice or combination is the correct one, it has profound implications on us.
4. It is possible to create and live in a simulated world, but not with the level of fidelity/flexibility necessary for the infinite regress that would make #3 convincing.