My earlier posts were intended to demonstrate that (2) is not possible: in order for it to be possible, we would have to be able to eventually simulate our entire universe, including every level of detail that we experience, which would not be even theoretically possible regardless to technology, because the computer that ran such a simulation would have to be bigger than the entire universe.
Nearly everyone's response to this is that base reality could be far larger and more complex, but that in itself means our reality does NOT resemble actual reality.
Bostrom's solution - that this is a sort of Truman show, that the sky is just painted on (and galaxies only resolve themselves once we have invented telescopes, and then only when someone is looking through them, that objects are solid until peered at though a microscope and only then do they become composed of microscopic phenomenon) doesn't really work either...
Take his example that the " structure of the inside of the Earth can be safely omitted" - for the beginning of the simulation that might work (depending how long its been running - it could have started with the dawn of life, or the dawn of man, the dawn of civilization, or it could have started 2 minutes ago, and your memories of this conversation were all pre-programmed).
Humans didn't have any expectation that the Earth was mostly liquid. That knowledge has had close to zero impact on anyone's behavior. Yet, once we discovered it was, now the simulation does have to consistently model the properties of molten rock swirling about a solid nickle core forever, because we have seismographs set up all over the Earth that can be cross referenced. It would take no less computing power to figure out what they would say only when they were actually crossreferenced than it would to actually just compute it in real time, so that doesn't save any computing power.
Or take neutrinos: their existence has not affected the path of human civilization in anyway (other than encouraging humans to build more neutrino detectors, and slightly modifying particle physics theory). They could have been safely omitted. And yet they are there.
It isn't enough to keep track of the mind state of individuals, the simulation would also have to keep track of when any detection equipment might randomly go off. Even if the simulation just fills in the data after the fact, it still has to compute individual particles that would have produced those results.
Objects in our reality actually behave as though they were made of particles, so a shortcut would change our experience.
All you need is a good magnifying glass to see Brownian Motion, yet in order for it to be simulated, every single molecule in the surrounding fluid needs to be individually calculated. But even without looking, fluids behave as though they were made of particles, so in order for things like weather to behave as if it were made of particles, you have to model all those particles.
Plus, in addition to simulating the environment we can see, it has to individually track all 7 billion humans to be aware if any of them might be about to look at a usually unseen phenomenon, AND it has to do enough calculations of every possibly observable phenomenon that IF anyone were to check, the results will be consistent with the rest of the world. Anything we might ever discover has to be tracked in advance, in order for it to be sure to be backward compatible with what we already know. This extra layer of complexity might take more computation and memory than actually just modeling everything to begin with.
If we are going to take his Truman Show explanation of the unseen not existing when it's not being looked, then what seems far more likely that there is only ONE consciousness in this simulation. That would be 7 billion times easier to track, and to keep the world consistent. Any one person is far less likely at any given moment to be off checking the granularity of reality, or how far the edges of the universe are. Chances are you (that on consciousness) have never even seen a neutrino detector, not even in pictures, so if its just you, and all the rest of us are stimulants, the simulator really could safely leave out neutrinos. And the motion and consistency of the center of the Earth. And individual protons and neutrons, and everything else which you trust other people (stimulants) are telling you the truth about.
This scenario is far far more plausible, yet seems to have next to no popularity. I suppose the idea that you are the sole conscious thing in the entire known universe just isn't as pleasant a thought as if we are all in this together.