Ask HN: Does simulation theory invalidate its own evidence?

The standard argument as I understand it goes: (1) Computing power grows exponentially -> (2) Civilizations will create ancestor simulations -> (3) Therefore we're statistically likely to be simulated

But if we're in a simulation, then (1) and (2) are just features programmed into our simulation, not facts about base reality. We'd be using potentially fictional evidence to prove the evidence is fictional.

It's like characters in a video game using the game's physics engine to prove they're in a video game - but the physics engine could be nothing like actual physics.

What am I missing? How do simulation proponents resolve this circularity?

Genuinely curious about responses to this, especially from those who find the simulation argument compelling.

[0] Bostrom - https://www.simulation-argument.com/simulation.html

[1] Moravec - https://www.faculty.umb.edu/gary_zabel/Courses/Bodies,%20Souls,%20and%20Robots/Texts/AI%20and%20Robotics/Simulation,%20Consciousness,%20Existence.htm

[2] Moravec article - https://www.wired.com/1995/10/moravec/

[3] Discussion that prompted this post: https://x.com/AndrewMayne/status/1953148275407913073