The simulation argument is trivially disprovable in many ways. For instance, look at all the hair and fur around you. It's too computationally expensive to simulate all this. Do you have any idea how many millions of dollars Pixar has spent simulating kind of shitty hair on like 20 characters?

@pietrokc I think hair is the least of the concerns of the simulators, but the underlying issue is that you'd need a world more complex than ours in order to simulate our world. Which is a thoroughly unsatisfying explanation, especially in the continued absence of anything other than anthropic reasoning supporting it.

@bens Yes, this is a bit. But it has a kernel of truth.

Of course you can believe you're in a simulation, just like you can believe in Descartes's malicious demon. But the well-known simulation argument -- whereby if we can simulate minds, we should conclude that we ourselves are simulated -- is trivially wrong for computational reasons.

If we can simulate minds, then the conclusion is that we live in a world where "depth-1" simulations are possible. Therefore, in order for us to conclude we ourselves are simulated, we have to postulate the existence (and in the original argument, high likelihood) of worlds in which "depth-2" simulations are possible. But that just does not follow from the existence of depth-1 worlds. Of course, if the minds we simulated themselves created simulations, we'd go to a depth-2 / depth-3 distinction, etc.

It is the same flavor of ignorance about how computers actually work that leads to the whole AI doom industry.

15%.

@Bayesian so low

You might think this is the kind of thing that's impossible to bet on, but I think it's doable.

If Bob's P(simulation) is <10% and Alice's P(simulation) is >90%, and they both trust each other to be intellectually honest, they can make a bet on their own opinions changing.

For example, they can make two bets:

A) $1000 to $1000 on if Alice's P(sim) will be <80% on January 1st 2028

B) $1000 to $1000 on if Bob's P(sim) will be >20% on January 1st 2028

This effectively becomes a bounty Alice and Bob to convince each other, but if they are good truth-seekers they should be quite happy to pay that bounty to double their credence on an important possible world state from 10% to 20%.

@Joshua if ppl care I can write a blog post on why serious physicists don’t really take the simulation hypothesis seriously. Because at this point, any non-falsifiable version of the hypothesis has been pretty soundly debunked.

@bens Surely you mean the falsifiable versions have been debunked?

🤔

@bens a lot of folks in the top half of this distribution… care to explain your reasoning?

@bens like, I don’t even think Nick Bostrom’s credence on this is above 90%, say.

@bens It's interesting that there are 7 votes for 1-10 and none for 90-99 but 2 for 99+