It's possible to run an experiment to see if a conscious entity could be convinced that a simulated reality was real. Hard to do ethically since consent would invalidate the illusion, but possible to demonstrate convincing virtual reality with computer generated sensory experiences.

Models like the holographic principle can be tested both mathematically and through experiment. Full tests are beyond our current capabilities but not unfalsifiable in theory.

>It's possible to run an experiment to see if a conscious entity could be convinced that a simulated reality was real.

This is like writing code to run a VM to detect if your code is being run inside a VM.

If the answer is no or yes it doesn't really say anything about whether the parent machine is running a VM with your code inside.

The results of the experiment wouldn't say anything about the simulation hypothesis.

This is a great example. You can prove that a VM can be just as capable as a bare metal OS. Then you connect to a random networked machine. Without any further information, would you guess that you chose bare metal or a VM (or container)? Knowing that virtualized OSs are both possible and heavily distributed, you would guess that any random machine you chose was probably virtualized.

The simulation hypothesis follows a similar logic with one less data point. Knowing that experienced reality could in theory be nested inside a superstructure but not knowing the actual deployment of such nested experiences, we would guess that our experienced laws of physics probably exist inside a superstructure.

What the properties of that superstructure are beyond the hypothesis itself because we don’t have the same knowledge of virtualization of physics as we do of OSs.

String theory is one attempt to describe that structure based on mathematical reasoning. The simulation hypothesis just states that physics is likely to be virtualized inside another system and that it’s worth exploring physics at its limits to understand the properties of that virtualization.

It’s falsifiable through the development of analytical methods that don’t fully exist yet but not theoretically unfalsifiable.

How "could in theory" suddenly became "likely"? The more contrived universe you imagine the less likely it is.

I think that's a misapplication of Occam's Razor. I'm not sure how exactly you're using contrived here, but the idea that complicated interactions like the ones we experience are likely to exist inside simpler systems is seeking the less contrived reason for existence. I'm saying that it's likely in the same way that virtual machines are likely to exist inside networks. Virtual machines might be more complicated than operating systems running on bare metal, but it's simpler and easier to create a thousand virtualized environments than one additional bare metal computer by orders of magnitude. In the same way, if it's possible to create nested experiencable universes, the total number of virtualized environments experienced is likely to be much greater than the number environments operating on the most bare, fundamental principles of existence alone.

This simulator's world is less likely, because it consumes more resources: needs to run thousand nested worlds, so their probabilities are reduced proportionally and further reduced by the virtual machine itself. Also bare metal is more efficient if virtual machine properly interprets instructions and doesn't merely forward them to processor.

Consuming more resources is a matter of perspective, not efficiency, because the scale of total phenomena would be much larger than what is experienced. Always bet against the comprehensiveness of your comprehension.