The logical explanation: "observation" has nothing to do with conscious woo, it's just that in order to have a definite answer we build experiments so they collapse the wavefunction.
It's like asking someone on a date: maybe they were in a superposition before, but now they have to answer, and having answered ("been observed"), that answer is highly likely to stay constant in the short term.
(when you think about it from this point of view, it's classical physics that's counterintuitive: why should we expect that asking questions about one projection of state doesn't affect the answers we get from later asking about others, not even in the slightest?)
The point I was trying to make is that if we are indeed in a simulation, and I'm not saying that we definitely are, but if we are - one possibility to design such a simulation in a way to make it more efficient is to actually make computations depend on the observer, meaning that sorry, but in this case it would have conscious "woo" built in.
Just in the same way as that only visible from current perspective objects are being drawn on a frame of a 3D game.
Currently unobserved parts of the simulation might exist in different form.
It's okay to disagree with simulation theory, but it is a perfectly valid possibility according to everything we know.
Personally, I don't think it's the only possibility, but i think it's quite probable and should be taken seriously.
One problem is that gravity is universally coupling, so no part of the universe is technically "unobserved." I suspect that we could look back at the dynamics of large scale systems and see deviations from GR if the simulation were neglecting any part of the universe in the absence of observation/interaction.
If I were building a simulation I would just have not made gravity universally coupling because it makes it hard to chunk reality up into parts. Thus it seems like the universal coupling of gravity is evidence against a simulation hypothesis.
The reason for my personal choice to not take simulation theory seriously is because simulations are an instance of Russell's Teapot. Anything which can be explained as S simulating T can be explained more simply as just T (or, in the opposite direction, even more complicatedly as R simulating S simulating T, etc. Can* we go all the way to a countably infinite tower of simulations?).
* if yes, then I'd have to admit that the omega-tower could be as interesting to study as the 0-tower, but if no then I'd maintain the 0-tower is way more interesting than any of its successor towers.
If you were to believe the universe was a simulation, would you do anything differently? could you?
(this line of approach, less formal and perhaps more congenial to CPS' other culture, is inspired by Dewey and James' pragmatism, in which philosophical problems are only well-posed if they have "cash value". We hackers don't make comparisons without subsequently using the flag value; they didn't ask questions whose answers are moot)
If we were in a simulation, due to the hard problem, it would be impossible for the simulators to know whether anything in their simulation had qualitative experiences, so they could not make conscious observation a prerequisite of detail rendering, only interaction. No woo necessary.
It's like asking someone on a date: maybe they were in a superposition before, but now they have to answer, and having answered ("been observed"), that answer is highly likely to stay constant in the short term.
(when you think about it from this point of view, it's classical physics that's counterintuitive: why should we expect that asking questions about one projection of state doesn't affect the answers we get from later asking about others, not even in the slightest?)
Does that make sense?