What happens to all the other people in the worlds the Waker leaves? I think I see five possibilities, all of which seem dreadful in one way or another.

• Their life goes on but the Waker has mysteriously vanished from it. In this case, a world with Wakers in sounds pretty bad. No one could rely on anything involving other people. (Unless Wakers chose to "wake up" extremely infrequently; say, only in circumstances like those in which someone in our world might choose suicide. But it seems clear that this isn't what you have in mind.)
• They are discarded, no longer being required. So we have, potentially, billions of people who exist only for the amusement of this Waker, and get turned off when the Waker loses interest. That seems morally dubious to me. I guess someone who has a problem with it wouldn't choose to be a Waker, but people change over time. How would it be to have a growing conviction that you're responsible for billions of deaths?
• They were never real in the first place, because they were only simulations. If you take the idea of being a Waker seriously, I think you have to reject this one out of hand. (So do I.)
• They were never real in the first place, because the Waker's computational substrate worked out exactly what other people would do without actually instantiating any. It seems unlikely that this is possible. At least some of those other people need to pass a very stringent version of the Turing test.
• They were never real in the first place, because the Waker found interactions with them plausible only because the computational substrate diddled their brain to make it seem that way. This seems basically indistinguishable from wireheading.

Any answer that begins "They were never real in the first place" (including those last three, but also any other possibilities I've overlooked) is open to the more general objection that a Waker's existence is (and is known to be) unreal and inauthentic and doesn't involve any interactions with actual other people.

[EDITED to add: Oh, I thought of another two possibilities but they're also awful:]

• They are immediately replaced by someone else exactly the same as them, so no one else even notices. Someone else exactly the same would do the same as they did, and someone else almost exactly the same would likely do the same very soon. Someone else exactly the same except suddenly deprived of the Waker's power? For that to work, it would need to be the case that Wakers might lose their powers at any moment, which seems like it would take most of the fun out of it. -- And if you do replace them with "someone else exactly the same", isn't that indistinguishable from saying that every time you attempt to Wake there's a 50% chance that it won't work?
• They are immediately replaced by someone else near enough for practical purposes. It seems like that would have to be very near, especially if there were other people to whom the Waker was very close. And a world in which people are constantly being created from scratch -- and put into situations someone else similar just decided they'd rather leave the world than deal with -- seems pretty unpleasant. As with the second option above, I think some Wakers would find themselves with big moral problems.

This is horrifying to me and I doubt that real posthumans exploring Fun Space will actually do this, just like they wouldn't eat every meal possible in order to explore Food Space. It's an unsophisticated human fetish.

Note that (2) and (3) are formal tasks where the optimizer has access to the full set of rules. My understanding is that a lot of chips designed by optimizing a simulator have been pretty lousy in the real world, either being complete failures that only worked in the simulator because of bugs, or being real solutions, but not being usefully robust.

Slime mold can be used to map subway routes.

Edit: Markets can also be seen as a non-human optimizing actor, even if the smallest parts are human.

Let's add here, that most of the scientists treat conferences as a form of vacation funded by academia or grant money. So there is a strong bias to find reasons for their necessity and/or benefits.

Well, glitch or not, I'm glad to have it; I would not want to be an unconscious automaton! As Socrates said, "The life which is unexamined is not worth living."

However, it remains to be seen whether consciousness is an automatic by-product of general intelligence. It could be the case that consciousness is an evolved trait of organic creatures with an implicit, inexact utility function. Perhaps a creature with an evolved sense of self and a desire for that self to continue to exist is more likely to produce offspring than one with no such sense of self. If this is the reason that we are conscious, then there is no reason to believe that an AGI will be conscious.

I agree; the OP is anthropomorphic; in fact, there is no reason to assume that an AGI paperclip maximizer would think like we do. In fact, in Superintelligence, Bostrom avoids any assumption that an AGI would have subjective conscious experiences. An unconscious AGI paperclip maximizer would presumably not be troubled by the fact that a paperclip is just an ill-defined configuration of matter, or by anything else, for that matter.

First everything in any practical simulation is always and everywhere an approximation. An exact method is an enormously stupid idea - a huge waste of resources.

We haven't seen anything like evidence that our laws of physics are only approximations at all. If we're in a simulation, this implies that with high probability either a) the laws of physics in the parent universe are not our own laws of physics (in which case the entire idea of ancestor simulations fails) or b) they are engaging in an extremely detailed simulation.

The optimal techniques only simulate down to the quantum level when a simulated scientist/observer actually does a quantum experiment. In an optimal simulated world, stuff literally only exists to the extent observers are observing or thinking about it.

And our simulating entities would be able to tell that someone was doing a deliberate experiment how?

The limits of optimal approximation appear to be linear in observer complexity - using output sensitive algorithms.

I'm not sure what you mean by this. Can you expand?

The upshot of these results is that one cannot make a detailed simulation of an object without using at least much resources as the object itself.

Ultra-detailed accurate simulations are only high value for quantum level phenomena. Once you have a good model of the quantum scale, you can percolate those results up to improve your nano-scale models, and then your micro-scale models, and then your milli-meter scale models, and so on.

Only up to a point. It is going to be for example very difficult to percolate up simulations from micro to milimeter for many issues, and the less detail in a simulation, the more likely that someone notices a statistical artifact in weakly simulated data.

We already can simulate entire planets using the tiny resources of today's machines. I myself have created several SOTA real-time planetary renderers back in the day.

Again, the statistical artifact problem comes up, especially when there are extremely subtle issues going on, such as the different (potential) behavior of neutrinos.

Your basic point that I may be overestimating the difficulty of simulations may be valid; since simulations don't explain the Great Filter for other reasons I discussed, this causes an update in the direction of us being in a simulation but doesn't really help explain the Great Filter much at all.