« Is it patronizing theft to buy natural resources? | Main | David Brooks, in a nutshell? »
Does the simulation have an evil or indifferent designer?
On the plane I was reading Stanislaw Lem's famous essay on personoretics. It occurred to me that if we are living in a simulation we can make Bayesian inferences about the intentions of the designer. Let's say many designers are creating many simulations. Will the good or the evil designers be more productive in terms of numbers of simulations created? If we define "good" as subject to some ethical constraints, I believe the good designers work under a competitive disadvantage. It's harder to produce cheap apples, for instance, if you pledge to do so only in a "green" way. And so on. Oddly the evil designers may be under a competitive disadvantage as well. Intention has a cost and so in competitive settings it tends to fall out. In our current world most things are made by indifferent machines. I believe the rational inference about the simulation is that at least the demi-urge -- and possibly the Master Creator as well -- is indifferent to our plight.
Posted by Tyler Cowen on May 14, 2008 at 05:38 PM in Philosophy | Permalink
Comments
Maximizing number of simulations seems like an unlikely optimality criterion. The simulations presumably have some goal besides existence, such as creating happy little software lives (which would select for "good" simulations), messing with things like a kid pulling off bug legs (selecting for evil simulations) or for research (favoring a wide array of goodness).
Posted by: Rick at May 14, 2008 6:49:42 PM
Are there any inferences to be drawn about the scope of the simulation?
Are we the ant farm on the table or a wasps nest three blocks over?
Posted by: burger flipper at May 14, 2008 6:52:36 PM
I hope Tyler is poking fun, if only slightly, at this whole simulation idea. That or he has been reading too much of Overcoming Bias. Personally, I find this an unlikely and unappealing answer to the problem of theodicy. I prefer the thoughts of John Stuart Mills and Thomas Malthus if you catch my drift. I wonder how many other early economists were part of the clergy...
Posted by: Sam at May 14, 2008 7:07:02 PM
But why are these designers creating the simulations? If the potential designers are just doing whatever they want to do, then indifferent designers are probably less motivated to create simulations than "good" and "evil" designers, which leads to the exact opposite prediction from Tyler's.
If both effects operate, that suggests that most simulations would come from non-perfectionist "good" and "evil" designers who are strongly motivated to produce simulations that are good or bad overall, but not especially concerned about producing high quality simulations that are uniformly heavenly or hellish.
The distribution of the motivations of potential designers is also important - we might expect "good" designers to be more common than "evil" ones, for instance, because why be evil when there's nothing in it for you? That's even more puzzling than altruism. If so, then it would be most likely for our simulation to come from a designer who is strongly motivated by good intentions, but somewhat sloppy about making things come out uniformly wonderful. Which, come to think of it, isn't all that inconsistent with what we see.
Wait - did I just finish an argument for the existence of a "god?" I think I did. Let me see if I can come up with any second thoughts... okay, got one. As anyone familiar with the past 400 years of productivity growth should know, simulations will be created much more efficiently by teams of "designers" who automate simulation production. That means that we are probably living in a simulation created by a machine produced by a group of designers. Yes, that must have a much higher prior probability than creation by an individual designer, and it also seems to fit even better with what we observe around us.
Posted by: Blar at May 14, 2008 7:31:35 PM
Can we make inferences about whether we have a can opener on the desert island? These seem equally valid to the simulation inferences Tyler makes. We can make any inference we want, but there are enough degrees of freedom that our inferences aren't any more useful than random guesses. Is our simulation created for artistic purposes, for some research purpose, or as a Monte Carlo process? We don't know. Furthermore, we can't make any useful guesses as to what the simulator would consider good art, what the Monte Carlo simulation (of simulations) would be approximating, and so on. The bottom line is that we have no way to conceptualize what indifferent, good, or evil intentions would mean and whether their costs would outweigh benefits.
Posted by: Greg at May 14, 2008 8:30:26 PM
Nice post and nice comment right above by this on by Greg.
I think we are probably one draw in a huge MCMC simulation being run by a grad student somewhere on Zorg who's writing her thesis on the evolutionary impossibility of individualism.
Posted by: angus at May 14, 2008 9:45:27 PM
If there is a competitive advantage to making simulations of people, then yes most sims will be those that are most competitive, which isn't obviously concerned about us. But possibly there is no competitive advantage to making sims, in which case sims are a side effect of slacker lazy silly whatever motives. And perhaps being good or evil comprise a large fraction of those motives.
Posted by: Robin Hanson at May 14, 2008 10:02:17 PM
If we had the ability to simulate the Earth in its earliest period, we could find out how life emerges, and the likelihood of other types of life. We might also vary the physical parameters, to try and evolve radically different types of life.
By analogy, the purpose of this simulated universe (the one we're in) is to understand the emergence of true (artificial) intelligence. Other universes might vary the population of researchers working on AI, in order to see what types of minds might emerge. The rest of the human race are just background.
The simulation ends when the AI emerges, and you no longer need the people.
Posted by: Michael Goodfellow at May 14, 2008 10:46:14 PM
I believe the rational inference about the simulation is that at least the demi-urge -- and possibly the Master Creator as well -- is indifferent to our plight.
I don't see how that follows. Why does the designer bother giving us rainbows, and the Pythagorean Theorem?
One could retort, "OK, then why did the designer give us torture and parallel parking?"
I think the right answer is, "Not exactly sure. It's obviously a lot more complicated than I realize."
But there's no way I could agree that the answer is, "The designer doesn't care about what happens to the homo sapiens put into the simulation."
Posted by: Bob Murphy at May 14, 2008 11:14:28 PM
Greg is right. In fact, what even justifies talking in terms of cost-benefit analysis? Anything powerful enough to create our entire universe could very likely be operating beyond any conception of scarcity. And, of course, if theres no concern for scarcity (in any sense of the word) then theres no sense talking in terms of ethics, either (well I sort of take that back - I suppose they could enforce them for arbitrarily, but in that case theres still no predicting their ethical choices).
Personally I don't rule out the simulation possibility. But if it is indeed the case for our particular universe, then there won't reasoning about its 'actual' nature. At least not without some sort of direct evidence.
Posted by: Leif at May 14, 2008 11:23:16 PM
sorry, the above should actually read 'there won't BE ANY reasoning' etc
Posted by: Leif at May 14, 2008 11:26:52 PM
If Borges were reading this he would propose a story in which an engineer, after years of labor, generates a simulated world, where she uploads some of her (perhaps dying) friends. She is proud of her work. By the end of the story, the engineer discovers that she is living in a simulation herself, and wonders about the creator of her world, and whether that creator is also simulated. She feels frustrated that she added one more futile link to an infinite sequence of simulated environments populated by simulations of simulations.
I do not know how the story would end. Unplugging?
Posted by: londenio at May 15, 2008 3:01:53 AM
Borges has no use for endings.
Posted by: Steve R at May 15, 2008 12:13:30 PM
Isn't theology fun?
Posted by: googaw at May 15, 2008 5:25:42 PM
Of course, Borges would speak of there eventually being some reality that isn't just a simulation in another world, but leave it open whether or not such exists. Still, "My solitude is gladdened by this elegant hope."
Posted by: Scott Scheule at May 15, 2008 5:31:03 PM
Does this mean I might be living in an incredibly sophisticated Civ game? So while I am playing Civ, a higher level being is playing the game in which I am "living"? Gosh
Posted by: techreseller at May 16, 2008 12:09:20 PM
Does this mean I might be living in an incredibly sophisticated Civ game? So while I am playing Civ, a higher level being is playing the game in which I am "living"? Gosh
Posted by: techreseller at May 16, 2008 12:09:49 PM
There will also be very large numbers of simulations produced during development and QA. They would run at some starting point until they go buggy.
So we're probably running in a flaky alpha version.
Posted by: John Gordon at May 18, 2008 7:59:18 PM
Last chaos Gold
Buy Last Chaos Gold
Cheap Last Chaos Gold
Posted by: aion kina at Mar 18, 2009 9:59:20 PM
