Friday, December 10, 2004

Living in Simulation

Here's a proposition (presented to me originally by Ben Hijink about a year ago) that some might argue challenges the limits of logical positivism:

Proposition One: We are all living in a computer simulation.

As a transhumanist, the idea of life in simulation is a familiar one. We could theoretically create intelligent life in a simulation (even if it's just a simulation of human cells), and control the universe of the simulated life as we wished. If we wanted, we could also guarantee that no experiment performed in the simulation ever gave the game away.

Star Trek: The Next Generation had an epidode entitled Elementary, My Dear Data that featured just this scenario. A computer simulation of Professor Moriarty becomes self-aware and eventually comes to learn that he is running in a simulation. Of course, Moriarty discovers his situation because the simulation isn't perfect (dang holodecks!).

Let's look at a refined version of the proposition:

Proposition Two: We are all living in a perfect computer simulation.

Here, perfection is in the sense that we can never find out that we are living in simulation on our own, and that the owners of the simulation will never reveal their secret. A logical positivist would say that this proposition is meaningless because no experiment the we ever perform will falsify, or confirm the proposition.

Yet, the creators of the simulation know that the proposition spoken by us is actually true because they built our universe. To the owners of the simulation, our proposition seems meaningful. You're probably beginning to see why people think this argument might have some relevance to the religious debate.

Can it be that logical positivism can claim that propositions are meaningLESS in sub-universes, but meaningFUL in enclosing universes?

Well, not exactly. While the creators of the simulation can say that Proposition Two is meaningful for the occupants of the simulation, they cannot say that Proposition Two is meaningful for themselves. The creators could themselves be living in perfect simulation!

The creators have built a mathematical system in which Proposition Two is meaningful, but outside of that system, the proposition remains meaningless.

Let's now return to Proposition One. Proposition One is very different than Proposition Two because it opens the possibility of transference of existence between simulation and the "real world."

Suppose we build our own simulated universe. In principle, we could see into the simulation we built, and we could give the occupants of that simulation a view of our world. For example, inside the simulation, the link to our universe might appear as a "magic" artifact. However, we can do better than this. We could create a portal into our universe. Upon passing through the portal, a simulant becomes wired into a robotic body, with the robotic sensors mapped to the simulant's senses.

Of course, we could also do the same in reverse, and transplant a person from our universe into the simulation. The idea of "uploading" people into computer systems is not implausible, and is frequently the topic of debate at transhumanist conferences. You just scan the person into digital form and simulate the action of their molecules.

We can now see that the premise of this scenario is that our intellect is independent of whether we are running on real atoms or on simulated ones. Our essence is information and computation, and the means of data processing is largely irrelevant. For the purposes of this debate, the barrier between the real world and the simulated world is not significantly different from a steel barrier between similar physical entities.

Clearly then, logical positivism must accept Proposition One as meaningful.

How does this affect the status of religious arguments under the principle of verifiability?

It doesn't. Proposition One is a purely scientific proposition. Propositions that explain UFO's in terms of alien spaceships are also scientific. Claims that bigfoot and the yeti are "missing links" in human evolution are scientific, too.

So, to the extent that a religious claim is a scientific, naturalistic claim, logical positivism is willing to accept it. This has always been the case.

However, most religions will never accept this. Their deities are beyond science and reason by definition. Such religious claims are meaningless because they insist that no experience can ever alter the probability of their being true. After all, who wants to worship the Klingons, even if their technology does look like magic to us?

Out of curiosity, is Proposition One metaphysical? Can a proposition be metaphysical and scientific at once?


Richard said...

Proposition One could be metaphysical or scientific or both depending on point of view. Or it could be undecidable in the Godel sense. Or it could be an example of the Church-Turing Halting Problem: "Any language containing conditionals and recursive function definitions which is powerful enough to program its own interpreter cannot be used to program its own 'terminates' function."
Witness the recent confusion surrounding the physics papers of the Bogdanov brothers and Dennis Overbye's
7 December article on string theory in The New York Times. Some critics suggest that string theory may be as much philosophy as science--that physicists are wasting their careers reading the entrails of goats.

Doctor Logic said...


I just read that NYT article. Since I left physics in 1993, string-theorists have made more progress than I had expected. I agree with Brian Greene that the replication of Hawkings result on black hole radiation is promising. String theory hasn't quite reached the philosophy stage yet.

I think that there is some connection between Proposition Two and Godel's theorem. When we create our own simulations, we artificially create a mathematical system in which certain questions are indeterminate and unaskable (i.e., Proposition Two is indeterminate inside the simulation). While we outside the system cannot make sense of the proposition in our own context, we can make sense of the proposition in the context of the simulation. This is very similar to Godel's theorem which applies to closed mathematical systems.

Perhaps Proposition Two is a sort of stealthy Set Theory problem that always resolves to a Godel-indeterminate proposition in any given mathematical system. If this is the case, we should be able to identify the mathematical structure of this proposition, then see if that same structure is present in other metaphysical propositions.


Nevin ":-)" said...

I don't see "We are all living a computer simulation" being any different than the proposition "We are all living in a G-d created world."

Those running inside the simulation would, as far as they can tell, have a different set of "physical laws" than those running the simulation. This is directly analogous to G-d being able to do things that us mere mortals cannot.

G-d could certainly open the possibility of transference of existence between his/her/its world and the "real world."

How can the question about the existence of G-d be meaningless, while the question about running in a simulation be meaningful?

Applying logical positivism: describe the experiment that would lend *any* evidence that we are running in a simulation. Or for that matter, describe the experiment that would lend *any* evidence that we are *not* running in a simulation?

It isn't that "logical positivism" breaks down; I just don't think that philosophy can be used to answer these types of questions. Which is of course what philosophers have said hundreds of years ago. Now, whether they were saying it to avoid persecution in their time, or whether they were saying it because they really believed it, is a whole 'nother question...

As a sidebar: statements like "most religions will never accept this" is, at best, a distraction. Whether or not a particular religion got the notion of G-d right is a very different (and not relevant) proposition from that of the existence of G-d.

Doctor Logic said...

Hi Nevin,

Glad to see you're keeping me honest!

You say that you don't see "We are all living a computer simulation" being any different than the proposition "We are all living in a G-d created world".

If the propositions are comparable, aren't you redefining god as an alien being that is subject to physical laws, albeit different from our own? And is that a religious proposition or is it a UFO proposition? I think what religion "accepts" is very relevant to our analysis.

Can there be scientific tests that would allow us to perceive the fact that we are living in simulation? Presumably, the owner of our simulation could have the ability to prevent us from ever knowing that we are living in simulation. It is also possible that the "sysop" might make it possible for us to obtain a scientific understanding of the universe which contains our simulation, revealing the nature of our simulation. Once we can break out of the simulation and identify the physical laws that govern the substrate of the simulation, we can gain a true scientific understanding of the simulation.

I'm not convinced that there's really any difference between a simulated universe and a "real" one (maybe superstrings are the computing mechanism). The way I see it, the proposition is a statement about different yet connected universes. The connection between the universes is that an entity in one universe created the second universe with foreknowledge of its simulative power. Within the sub-universe, interactions between the universes may be 1) imperceptible, 2) perceptible with insufficient structure to be modeled, or 3) perceptible with sufficient structure to be modeled. The simulation proposition is a claim about scenario 3). It cannot be about scenario 2) because scenario 2) is indistinguishable from a great many other (semantically different) propositions that would produce phenomena that cannot be modeled.

You talked about the question of the existence of god. Can a thing exist independent of the sense data corresponding to that thing? Does your TV set exist independently of any of its measurable properties? When we talk about the existence of a thing, we are using a shorthand for the possibility that we will measure the conjunction of observables that define the named thing. We use this shorthand so casually that we can easily make the mistake of thinking that existence is independent of measurable properties. My claim is that to speak of god's existence (in the traditional sense) without reference to his measurable properties is neither true nor false, but nonsensical because it misappropriates the word "existence".


Doctor Logic said...

BTW, Nevin, thanks for challenging my reasoning.

I don't have all the answers (at least not pre-computed, as far as I know!), so each challenge requires me to probe the structure of positivist ideas in new ways.


Anonymous said...

It all depends upon what you consider to be a "simulation". A simulated world follows a set of well-defined rules, and similarly so does the real world. The only difference really is the complexity and subtlety of those rules.