We know that systems can be nested. Do we know that the systems we perceive and can test directly compose the most fundamental layer? How can we tell? If we want to have any faith in science and testing our direct experiences, we need to keep those questions in mind.
Can a world be simulated in a computer?
If yes, how many worlds and individuals are simulated at the same time? (And it may make sense for example market research, science etc. Of cause the simulation will run much faster then the real world).
Then there is a problem. If the world CAN be simulated and MUCH MORE simulated individuals exists than real persons it is more likely for you to live in a simulation than in the real world. A point worth thinking about.
So in the event that someone holds a firm an unwavering belief that we're in a simulation unrelated to availability of facts, then their belief would be akin to a religious belief, sure, but the intent of the simulation argument is not to claim that there is evidence we live in a simulation.
Specifically, it does not argue that we live in a simulation at all.
It presents the hypothesis that at least one of the three given propositions must be true (paraphrased):
1. Either very few human-level civilizations reach a stage where they can run ancestor simulations, OR
2. Very few such civilizations have any interest in running such simulations, OR
3. Most people live in a simulation.
Basically it boils down to an argument that if many civilizations reach a stage where they can and do run such simulations, then our odds of living in simulation by pure chance will be high.
But the argument says nothing about whether or not we live in a simulation in isolation, because the argument does not quantify the odds of 1 or 2, and even if it happens that the chances of living in a simulation are astronomically high chance, that is still not evidence that we are, it would just be an indication that we probably are.
But let's say it turns out we have something akin to a "soul" that can't be simulated. In that case the number of civilizations in #1 is 0, and we're not in a sinulation under any circumstances. Or the odds of hitting a Great Filter and die out before we reach the technological level necessary might simply be insanely high, in which case the number in #1 might very well be very close to 0.
It may also be that any civilization that reach such a stage first will hit a singularity or other event that changes its priorities such that nobody are interested in such a simulation, or can justify the effort or expense. In which case the number of civilizations interested in running such simulations even if possible may be close to 0 (these options may also affect #1 - it may be we opt to change our future in ways that sends us off in a different technological direction).
The point of the argument is exactly the questions it raises, first and foremost by making people think about the possibility, and secondly to make people think about what may prevent us from being able to (which is tangentially related to the Fermi paradox), and what might make us want to or not want to, and lastly the questions it raises about whether there are ways for us to quantify or investigate each of the propositions. E.g. if #3 holds, are there any ways we can determine if we are in a simulation or not?
It is a philosophical argument, put forth to present the possibility that we live in a simulation that similar to e.g. the Drake equation seeks to define a set of parameters to focus a discussion around. It is not the intent that we should just assert the 3rd proposition is true, as even if it is true, it does not resolve all the questions around the 1st and 2nd, and it would just raise new question (the hypothesis can apply recursively - if it is possible to run such a simulation, then if it is posible to also run such a simulation within a simulation, then how that does affect the propabilities? it might seem as it creates a rather large subset of possible solutions that would make the probability of being in an infinitely nested simulation approach infinity; is there something that would prevent that?)
So it's not a bad scientific theory - it's a good philosophical hypothesis. It might in the future give rise to scientific theories. E.g. attempts to define falsifiable bounds to the probabilities, or falsifiable predictions about properties the world must have if we are in a simulation or not in one. But for the moment, it is a philosophical argument, and attacking it for not being a scientific theory is aiming in the wrong direction.