NAWKI November 2017 - Living in a simulation

01 Feb 2018

NAWKI (life, but Not As We Know It), is a monthly series exploring the hypotheticals of real and artificial life


pennyfaulkner: So, I’m going to jump the gun and announce the next #nawki for November: “What if we were all living in a simulation (and we knew it)?” I want to talk about this since this article popped up in my news feed this morning.

It’s a pretty nice summary of the common arguements relating to a simulation we don’t know about and raises for me an interesting question. In this theoretical set up, how do we know we’re in a simulation?

Surely we get into “indistinguishable from magic” territory here. If we know there is some great controller observing and manipulating our world wouldn’t we just call it god? How is living in a simulation distinguishable from having absolute knowledge of an intelligent creator?

d9w: I don’t think it is, except that we have no verification of the creator. Personally, that is an argument against the simulation hypothesis, for me: why does a simulation need to be deceptive to its inhabitants to make them unaware of the fact that they are in a simulation?

It seems like there’s some consensus that, due to the non-computable vastness of the universe, a simulation would only compute what is observed in the universe. The creator then has to be aware of what is being observed in their simulation, and actively increase computing to continue to deceive the observers

That falls in line well with Bostrom’s imaging of a simulated universe, because those are mostly about ancestor simulations: a superintelligent human race that is replaying parts of history. For me, that seems like a very narrow slice of computer simulations that could involve sentient beings in a seemingly physical universe.

I haven’t reached anything close to simulated sentience, but I do run simulations of beings with artificial brains in a world with artificial physics. They look nothing like human ancestors, nor do I concern myself with the realism of their observations

As for the question of, “if we knew it”, I think there are two different things here: If we abstractly knew that the universe was a simulation, but not why or how, I assume it wouldn’t make much difference to most people, and there would be some who pursue the purpose and creator. I’d argue that’s the state we’re in: there are people actively trying to figure out how to get out of the simulation, if we’re in one.

If, however, we were aware of the purpose and maybe a creator, that, to me, would look like religion. You try to abide according to the principles that the creator dictates, and even try communicating with the creator.

This is of that mindset, and argues that we should live to be famous and entertaining, because that’s what our super-intelligent descendants will want to replay. However, this is again in the vein of ancestor simulations, and if we are living in a simulation, I personally doubt it’s just a historical playback.

pennyfaulkner: I feel only using simulations for historical playback show a distinct lack of imagination. Why not simulate alien life or alternate history. Way more entertaining.

d9w: Agreed! And yet the source for this, Nick Bostrom’s paper, seems to consider this the main type of simulation.

Iñaki Fernández Pérez: Totally agree with you both: with enough computational power to do historical playback, why not simulate something more interesting such as “life but not as we know it” (either in alien worlds, imaginary worlds, or an alternative history on earth’s events).

That said, the argument about religion seems particularly interesting to me. If we are living a simulation, or if such simulations exist, there is indeed something or somebody that built such simulation, a “creator”. I prefer to write “creator” in quotes to avoid misunderstanding. Following our hypothesis, such a “creator” must have knowledge of all the laws that govern the simulation (or at least those whose consequences we may observe), and that simulation must be executed on a computational substrate of equal or higher complexity.

Assuming as second hypothesis that we had knowledge that we are in such a simulation (for whatever means necessary), the consequences, according to the bbc news article, may be non-existing for our lives. In this sense, we could think of our “creator” as the Creator (in the religious sense), and that is where things start to get messy. However, in terms of what is computationally tractable for such a simulation, in my opinion we probably could not know the details on how that simulation works (or how the “creator” created the simulation). The reason stems by the fact that if we knew how to create an exactly similar (second-order) simulation, we would have created it, representing a complex universe, where “sentient” entities could also have access to such knowledge, which would extremely increase the computational requirements for a (first-order) simulation to run, and this argument could be applied recursively. This remains an opinion and I do not have stronger arguments to support it, but the fact that in this hypothetical situation we could know that we are in a simulation, but not how the simulation operates, claims for a religious view, or for the “indistiguishable from magic” territory that @pennyfaulkner mentioned. As for the “playback” argument of the paper @d9w provided (not Bostrom’s but the other), it just feels sketchy. If I am a superentity with extreme “power”, why would I stop at just replaying history as it was, and not use it for simulating alternate history or something similar, which could be helpful for my “supersociety”. Now that I think about it, the reason could be that other “superentities” enjoy viewing simulations as entertainment, but this is just plain digression

comments powered by Disqus