> Some don’t adopt the self-indication assumption (SIA). The most common alternative is the self-sampling assumption (SSA), according to which one should, upon discovering the splendid fact that they exist, reason as if they’re randomly selected from all the people like them. Thus, you should think there probably aren’t lots of people like you on Neptune, for instance, because if there were then it would be quite odd that you ended up on Earth rather than Neptune.
I haven’t finished the article yet but my understanding is that this is slightly incorrect. SIA and SSA are not alternatives, they are orthogonal to one another.
SSA forms a bridge that connects theories proposing a distribution of observers, and hypotheses predicting with hypotheses predicting what someone should exist to see if such theories were accurate.
SIA on the other hand informs our priors about those theories, causing us to prefer theories that have more observers.
There is no reason we can’t have both, and there are some reasons to. When choosing between multiple hypotheses with different numbers of observers when you’re in a group that would exist in either case (an often confusing scenario in anthropics, eg doomsday), SIA and SSA cancel each other out. Having both also removes the need for the poorly understood concept of a “reference class” of observers, as the number of observers appears both in the numerator and denominator of most judgements and so becomes arbitrary.
Yeah it’s used in different ways in the literature. Bostroms formulation is a bit idiosyncratic but what I mean by sia is what bostrom means by ssa+sia.
Are you saying that the number of people is of a literally indescribably magnitude? Is there no mathematical conception that can contain the number of people?
1. Doesn't the discreet discriminability and therefore countability of possible require more underlying reasoning. My objection is going to be hilariously undercomplex compared to your stacking of Beth's, but I'm reminded of a finite dartboard with infinite possible locations for the dart to land at any one of them, all of them with probability of zero. In that scenario, anything regarding probabilities would always involve areas instead of points.
2. The dartboard-objection loses it's (possibly non-existant) power once physicality enters the mix, because very weird things happen when the granularity of points increases so much that quantum effects begin to mess everything up, such that you probably can't keep going to stack more and more infinities on top of each other just by counting points. However, (possibly entirely unrelated), the fact that computation on a silicon die is a theoretically energy-neutral operation if it weren't for waste heat - meaning the electrons don't really do something, unlike every other type of electrical appliance - give some strong evidence that physical limits don't apply the same way when we're in the realm of information processing, which minds would be assigned to (at least I would think), since the physical side of information processing only ever acts as a state-representative proxy instead of the actual thing.
> Some don’t adopt the self-indication assumption (SIA). The most common alternative is the self-sampling assumption (SSA), according to which one should, upon discovering the splendid fact that they exist, reason as if they’re randomly selected from all the people like them. Thus, you should think there probably aren’t lots of people like you on Neptune, for instance, because if there were then it would be quite odd that you ended up on Earth rather than Neptune.
I haven’t finished the article yet but my understanding is that this is slightly incorrect. SIA and SSA are not alternatives, they are orthogonal to one another.
SSA forms a bridge that connects theories proposing a distribution of observers, and hypotheses predicting with hypotheses predicting what someone should exist to see if such theories were accurate.
SIA on the other hand informs our priors about those theories, causing us to prefer theories that have more observers.
There is no reason we can’t have both, and there are some reasons to. When choosing between multiple hypotheses with different numbers of observers when you’re in a group that would exist in either case (an often confusing scenario in anthropics, eg doomsday), SIA and SSA cancel each other out. Having both also removes the need for the poorly understood concept of a “reference class” of observers, as the number of observers appears both in the numerator and denominator of most judgements and so becomes arbitrary.
You are wrong. Read bostroms book, for example.
Here’s a presentation from Bostrom where he says the same thing I said: https://youtu.be/oinR1jrTfrA?si=yjtM1KmnFT_sveQZ
Yeah it’s used in different ways in the literature. Bostroms formulation is a bit idiosyncratic but what I mean by sia is what bostrom means by ssa+sia.
So when others Eg carlsmith are talking about sia they mean something roughly like what bostrom refers to as sia+ssa.
Sorry I realize my original comment was pretty abrasive! Didn’t intend it to be so!
The relevant part starts 26 minutes in
Ok Bulldog. Assuming you’re right, how many people exist?
More than can be part of any set.
Unhelpful. I could say “more than one”.
Are you saying that the number of people is of a literally indescribably magnitude? Is there no mathematical conception that can contain the number of people?
So there are infinite sets. Some infinites are too big to be a set—eg there’s no set of all truths. The number of people is like that.
1. Doesn't the discreet discriminability and therefore countability of possible require more underlying reasoning. My objection is going to be hilariously undercomplex compared to your stacking of Beth's, but I'm reminded of a finite dartboard with infinite possible locations for the dart to land at any one of them, all of them with probability of zero. In that scenario, anything regarding probabilities would always involve areas instead of points.
2. The dartboard-objection loses it's (possibly non-existant) power once physicality enters the mix, because very weird things happen when the granularity of points increases so much that quantum effects begin to mess everything up, such that you probably can't keep going to stack more and more infinities on top of each other just by counting points. However, (possibly entirely unrelated), the fact that computation on a silicon die is a theoretically energy-neutral operation if it weren't for waste heat - meaning the electrons don't really do something, unlike every other type of electrical appliance - give some strong evidence that physical limits don't apply the same way when we're in the realm of information processing, which minds would be assigned to (at least I would think), since the physical side of information processing only ever acts as a state-representative proxy instead of the actual thing.