science

. That is, the parent universe which is running our simulation must run it for an impracticably long time.

I understood that. I'm pointing out that you're making an assumption that trillions of years is an 'impractically long time'. It is to us, but there's no reason it would be to another universe. Assuming time even works the same way and isn't just a cool thing they came up with for this simulation.

yes each simulation has to be a similar universe to the one we ourselves live in. Only that way do you end up with vastly more simulated universes than real universes, and the conclusion that statistically we must be living in a simulated universe and not a real one.

Firstly, the current discussion isn't about the probability of us living in a simulation. It's about whether it's possible to begin with.

Secondly, 'each simulation has to be a similar universe to the one we ourselves live in. Only that way do you end up with vastly more simulated universes than real universes' is in itself another assumption that doesn't necessarily hold true. The only thing the simulated universes need to have in common is that they contain sentience intelligent enough to continue the chain of nested simulations. The physical rules governing each simulation might be wildly different.