Are We Living In a Computer Simulation?

[anhinga_anhinga]

Cema pointed my attention to the fact that one of the smartest modern philosophers, Nick Bostrom, now has a separate Web site dedicated to the Simulation argument.

Comments

[anhinga-anhinga.livejournal.com]

On the other hand, various recursive models are of interest here ;-)

One of my favorites is that we are created by our post-human descendants ;-)

[cema.livejournal.com]

This is good for sci fi, but not good enough for specs. I want specs.

[anhinga-anhinga.livejournal.com]

Having proper specs here is more or less equivalent to being post-human, I think.

There is an informal notion of an AI-complete problem. Having the correct theory of the world is probably at least AI-complete. And having decent specs here is unlikely without a much better theory of the world, than we have now...

We should pick a somewhat less radical project for making specs...

(BTW, how would you describe the difference between specs and implementation?)

[cema.livejournal.com]

I will start from the end. The specs are good for a human being. When this human being can convince the computer to behave according to the specs, that's implementation. Therefore, if the computer is as intelligent as a human, there is no difference.

I do not see, however, why in order to implement the simulation one would need to come through the singularity. Furthermore, I do not see why coming through the singularity would necessarily lead to the simulation becoming possible.

Finally, a "correct theory of the world" is, strictly speaking, impossible, as follows from the trivial observation that, in order to simulate a universe, one would have to create its copy (another universe), which would be futile anyway, because no measurement would be possible (it would destroy the similarity between the universes). So, instead of a correct theory, we need something more flexible: a plausible theory. Now this is interesting, but I did not think sufficiently hard about this to post anything, even in a comment. (Maybe we will have time to discuss this tomorrow?)

[anhinga-anhinga.livejournal.com]

>I do not see, however, why in order to implement the simulation one would need to come through the singularity. Furthermore, I do not see why coming through the singularity would necessarily lead to the simulation becoming possible.

Mostly, because the simulation of this kind is AI.

>correct theory

I mean, a theory based on "correct" principles. Of course, we can only talk about "more correct" and "less correct", like in "general relativity is more correct than the preceding theory of gravity".

Basically, I don't think the existing understanding of the world is "correct enough" to tell us what is required to create an experience of a human-like life by running a computer program.

>a plausible theory [...] Maybe we will have time to discuss this tomorrow?

Indeed.