>I know I have some particular set of experiences. If there were fewer people, then while there might be someone having the experiences I’m having, it’s less likely that my particular experiences would be had by me.
Isn't this false in general, and thus in need of greater qualification? Suppose God flips a coin: if heads, he creates a thousand people and randomly assigns ten a red jacket and the rest blue jackets; if tails, he just creates one person with a red jacket. I notice my jacket is red. This is obviously going to be evidence for the tails outcome (since observing blue would be definitive evidence for the heads outcome), in spite of the fact that fewer people with my experience of seeing a red jacket exist on tails.
But your existence is a necessary fact. It is metaphysically impossible for you not to exist, because the act of calculating the probability of your own existence comes only after your own existence has been established.
This is entirely distinguishable from contraception because contraception doesn’t impact the probability of whether you exist, it impacts the probability of you being a flesh and blood human whose body was created by sexual reproduction. The changes of you specifically existing in some form are always 100%.
Saying that it’s about your own experiences is also not relevant. If you exist, maybe you will have some arbitrary set of experiences, but since you’re guaranteed to exist, having some arbitrary set of experiences doesn’t predict anything. The chances of you having this are 100% either way.
> Here’s one way that makes good sense if you believe in souls
I agree that SIA makes sense if souls existed, but this is not a point in favor of SIA. If you require to postulate non-physical entities to make sense of your probability theory application, it seems that you've made a few wrong steps along the way.
> Suppose that I’m considering two hypotheses. Both predict that there are two people in a room. One of them predicts that both of them will, when they look under a table see a red strip of tape, while the other theory predicts that one of the two people will see a red strip and the other a blue strip.
This is a great example. Let's explore it in more details.
Suppose that you are none of these people. However, there is some mechanism that tells you about what one of these people observed. Let's say that there is a red light in your room. And you know that if the light goes on it means that a red strip was observed. But you are not sure about the actual algorithm. Does the red light just always goes on because there is always a person who observes a red strip? Or does it sometimes goes on because of it, and sometimes it doesn't, regardless of whether there are two red strips or not? Or does the red light correspond to the particular person and specifically when this person observes the red strip it goes on and when this person doesn't observe a red strip it doesn't? If it's the latter, can this person possibly not observe the red strip? Are strips assigned to people at random and a random person observation of the red strip leads to red light going on? Or is the person who gets assigned the red strips is also always the person whose observation of it leads to turning the red light on?
In other words, either observation of red light is not correlated with any of two hyposthesises or it is. You've observed one of these events:
1. "At least one person has observed a red strip" or you observe event
2. "A person who always observe a red strip has observed a red strip"
3. "A person who could have not observed a red strip has observed a red strip"
According to conservation of expected evidence you are not supposed to update in 1 and 2 but you are supposed to update in 3.
The problem is that you can't a priori deduce whether you observed 1, 2 or 3. The same universe that allows you to observe 3 can just as well let you observe 1 and 2. And so to simply commit to treat all the situations the same way, without inquiring about the specifics of the causal mechanism is a clear mistake. And this is why SIA is wrong in a general case.
> Upon finding this out that the raspberry exists, you get evidence for theories on which there are more raspberries.
Suppose that there are two theories. That there are 100 raspberries in a bag or that there are 1000 raspberries in a bag. You are given a particular raspberry from a bag. You don't actually update in favor of there being more raspberries in a bag, do you?
Does SIA suggest that I should accept a form of panpsychism where the universe is packed with consciousnesses experiencing exactly what I'm experiencing? It seems to suggest a vast infinity of consciousnesses exactly like mine, perhaps one at every point in space.
(It seems like the problem of trapped priors.) There's no reason to assume anything if you wake up in a room. Sure, you might wake up with crazy beliefs (priors) and then you can conclude anything. But that's not a great argument for this (meta)strategy.
> alternatives to SIA don’t really treat your existence as something that you update on
Which distribution you would need to update? Was your existence so underestimated previously, that you need to update on it? Your existence is already part of your prior(s), and when you notice it, there's no delta. No new information.
When you look around you update on the existence of things you see, that you can move your head, the experience of changing viewpoint (moving).
And sure, experiencing things should make us consider that there might be other beings with experience (but we don't know anything about how similar/different theirs might be).
And thus upon encountering strange questions about possible worlds ... we can use a Bayesian approach, but our prior is consistent with both versions to exactly the same degree, hence we have to conclude that both are equally likely theorems/scenarios. (This might be the only room, we might be the randomly picked lone A, and so on. [And it doesn't matter if we know that if God were to iterate the setup we would exist in 5/6 of the cases, because we don't have data about this. Sure we might be living in a simulation, if we assume a world where that's likely. But that's just failing Ochkam.])
I think you're very right that, if you have some sort of essentialist model of personhood wherein everyone is most basically a soul or something similar, then a lot of metaphysical objections to the SIA are not problematic. But if you don't take that view, then I think there's a plausible story for the SSA that makes a lot of sense. I also think the concern about reference classes is overblown - it doesn't seem weird to me that we would be unsure about whether a particular entity is someone "you could have been," and it might even be an empirical question.
Finally, I'd just say that there is a sense in which the SSA'er updates on their existence, right? When you find that you exist, you narrow down the worlds you're considering to those in which you exist and you don't consider the ones in which you don't. That's the narrative that makes sense to me. You're basically saying, "Hey, it turns out I exist, so however many people exist, I'm one of them" and then you go on to think about all the different possible worlds where that's true. This is why I don't get the "why do you think you would be created in any world?" objection - it's not that I think I'm guaranteed to exist in any world that gets created, but just that, from my own existence, I know those worlds can't possibly be actual so why would I consider them?
(Also, Joe Carlsmith has a sorta funny mystical explanation for the SSA in his big piece, if you remember, and honestly I think someone could take that and run with it if they really wanted to because 1) it's kinda cool and 2) I still think it's more plausible than infinitely many people actually existing.)
>I know I have some particular set of experiences. If there were fewer people, then while there might be someone having the experiences I’m having, it’s less likely that my particular experiences would be had by me.
Isn't this false in general, and thus in need of greater qualification? Suppose God flips a coin: if heads, he creates a thousand people and randomly assigns ten a red jacket and the rest blue jackets; if tails, he just creates one person with a red jacket. I notice my jacket is red. This is obviously going to be evidence for the tails outcome (since observing blue would be definitive evidence for the heads outcome), in spite of the fact that fewer people with my experience of seeing a red jacket exist on tails.
But your existence is a necessary fact. It is metaphysically impossible for you not to exist, because the act of calculating the probability of your own existence comes only after your own existence has been established.
This is entirely distinguishable from contraception because contraception doesn’t impact the probability of whether you exist, it impacts the probability of you being a flesh and blood human whose body was created by sexual reproduction. The changes of you specifically existing in some form are always 100%.
Saying that it’s about your own experiences is also not relevant. If you exist, maybe you will have some arbitrary set of experiences, but since you’re guaranteed to exist, having some arbitrary set of experiences doesn’t predict anything. The chances of you having this are 100% either way.
> Here’s one way that makes good sense if you believe in souls
I agree that SIA makes sense if souls existed, but this is not a point in favor of SIA. If you require to postulate non-physical entities to make sense of your probability theory application, it seems that you've made a few wrong steps along the way.
> Suppose that I’m considering two hypotheses. Both predict that there are two people in a room. One of them predicts that both of them will, when they look under a table see a red strip of tape, while the other theory predicts that one of the two people will see a red strip and the other a blue strip.
This is a great example. Let's explore it in more details.
Suppose that you are none of these people. However, there is some mechanism that tells you about what one of these people observed. Let's say that there is a red light in your room. And you know that if the light goes on it means that a red strip was observed. But you are not sure about the actual algorithm. Does the red light just always goes on because there is always a person who observes a red strip? Or does it sometimes goes on because of it, and sometimes it doesn't, regardless of whether there are two red strips or not? Or does the red light correspond to the particular person and specifically when this person observes the red strip it goes on and when this person doesn't observe a red strip it doesn't? If it's the latter, can this person possibly not observe the red strip? Are strips assigned to people at random and a random person observation of the red strip leads to red light going on? Or is the person who gets assigned the red strips is also always the person whose observation of it leads to turning the red light on?
In other words, either observation of red light is not correlated with any of two hyposthesises or it is. You've observed one of these events:
1. "At least one person has observed a red strip" or you observe event
2. "A person who always observe a red strip has observed a red strip"
3. "A person who could have not observed a red strip has observed a red strip"
According to conservation of expected evidence you are not supposed to update in 1 and 2 but you are supposed to update in 3.
The problem is that you can't a priori deduce whether you observed 1, 2 or 3. The same universe that allows you to observe 3 can just as well let you observe 1 and 2. And so to simply commit to treat all the situations the same way, without inquiring about the specifics of the causal mechanism is a clear mistake. And this is why SIA is wrong in a general case.
> Upon finding this out that the raspberry exists, you get evidence for theories on which there are more raspberries.
Suppose that there are two theories. That there are 100 raspberries in a bag or that there are 1000 raspberries in a bag. You are given a particular raspberry from a bag. You don't actually update in favor of there being more raspberries in a bag, do you?
Does SIA suggest that I should accept a form of panpsychism where the universe is packed with consciousnesses experiencing exactly what I'm experiencing? It seems to suggest a vast infinity of consciousnesses exactly like mine, perhaps one at every point in space.
(It seems like the problem of trapped priors.) There's no reason to assume anything if you wake up in a room. Sure, you might wake up with crazy beliefs (priors) and then you can conclude anything. But that's not a great argument for this (meta)strategy.
> alternatives to SIA don’t really treat your existence as something that you update on
Which distribution you would need to update? Was your existence so underestimated previously, that you need to update on it? Your existence is already part of your prior(s), and when you notice it, there's no delta. No new information.
When you look around you update on the existence of things you see, that you can move your head, the experience of changing viewpoint (moving).
And sure, experiencing things should make us consider that there might be other beings with experience (but we don't know anything about how similar/different theirs might be).
And thus upon encountering strange questions about possible worlds ... we can use a Bayesian approach, but our prior is consistent with both versions to exactly the same degree, hence we have to conclude that both are equally likely theorems/scenarios. (This might be the only room, we might be the randomly picked lone A, and so on. [And it doesn't matter if we know that if God were to iterate the setup we would exist in 5/6 of the cases, because we don't have data about this. Sure we might be living in a simulation, if we assume a world where that's likely. But that's just failing Ochkam.])
The fact that there are people who deny SIA shows that some philosopher or other will defend any kind of insane view.
I think you're very right that, if you have some sort of essentialist model of personhood wherein everyone is most basically a soul or something similar, then a lot of metaphysical objections to the SIA are not problematic. But if you don't take that view, then I think there's a plausible story for the SSA that makes a lot of sense. I also think the concern about reference classes is overblown - it doesn't seem weird to me that we would be unsure about whether a particular entity is someone "you could have been," and it might even be an empirical question.
Finally, I'd just say that there is a sense in which the SSA'er updates on their existence, right? When you find that you exist, you narrow down the worlds you're considering to those in which you exist and you don't consider the ones in which you don't. That's the narrative that makes sense to me. You're basically saying, "Hey, it turns out I exist, so however many people exist, I'm one of them" and then you go on to think about all the different possible worlds where that's true. This is why I don't get the "why do you think you would be created in any world?" objection - it's not that I think I'm guaranteed to exist in any world that gets created, but just that, from my own existence, I know those worlds can't possibly be actual so why would I consider them?
(Also, Joe Carlsmith has a sorta funny mystical explanation for the SSA in his big piece, if you remember, and honestly I think someone could take that and run with it if they really wanted to because 1) it's kinda cool and 2) I still think it's more plausible than infinitely many people actually existing.)