I think that Carroll’s argument is wrong because it already assumes as true what it wants to disprove.
His argument is the following claim: If I am BB, I can’t reason about physics and the probabilities of being BB. However, this claim is true only if I am BB. But if I am interested only in the question “Am I BB?”, it doesn’t matter how good I am thinking of being a BB or everything else, because the only thing which I am interested in is already assumed as true.
The same logic was used in the attempts to disprove the simulation argument by Danila Medvedev (and maybe others). If I am in a simulation, I can’t have a perfect model of the distribution of simulations in the universe and—Danila suggests, – it disproves the simulation argument.
Let’s elaborate: One of two statements is true:
I am real and simulations are impossible.
I am in simulation, but in that case, I can’t perform perfect probability estimations.
The italic part doesn’t have any logical power over the whole argument.
Carroll argument resembles a typical proof by contradiction: let’s assume A, but it produces a contradiction, so A is false. However, in Carroll’s case, no matter what is happening with BB, there is no contradiction, so it is not a proof that I am not BB. This is even stronger in the case when it is applied to the simulation argument, as a sim can think coherently.
But what Carroll actually said is that BB can’t know for sure that it is BB.
But from my point of view, its all is compensated by the dust theory: for any BB1 there is a BB2 which looks like logical continuation of BB1. Like for any random integer N there is another integer N+1.
Therefore, for any BB which sees absurdity, there is another BB which recognise it as absurdity. Thus in dust theory, chains of BBs can calculate probability that they are BB.
Therefore, for any BB which sees absurdity, there is another BB which recognise it as absurdity. Thus in dust theory, chains of BBs can calculate probability that they are BB.
If you are a BB at all, you have a vanishing likelihood of being a sane BB.
Thanks for the link to Wallace’s work. I think that he also fails in the same trap, that is he fails to distinguish “large” and “small” forms of self-undermining arguments.
The large one is a choice between two alternatives: “1) I am real; 2) I am BB, and therefore can’t have reasonable thoughts.”
The small argument is only (2) from above: “I am BB, and therefore I can’t have coherent thoughts. “Only the small argument is self-undermining but the large one stands because there is no probability update in (1) horn.
For example, I have a 10 per cent chance to believe that I am real a priory, based on theoretical physics. I know that BBs can’t update properly but it is not evidence on which I should update, as it is not news for me.
Another example: I can be either awake or dreaming. During dreaming I can’t effectively decide, am I awake or dreaming and properly calculate the chances of being in a dream. Sometimes I have lucid dreams and sometimes I have very realistic dreams. If we follow Carroll’s logic, we should conclude that dreaming is impossible, because it is cognitively unstable, and it could be that I am dreaming now about commenting in LW. But in real life, we assume that dreaming takes, say, 0.3 time of my subjective experiences and we ignore any probability computations about dreaming that we do during dreaming.
I think that Carroll’s argument is wrong because it already assumes as true what it wants to disprove.
His argument is the following claim: If I am BB, I can’t reason about physics and the probabilities of being BB. However, this claim is true only if I am BB. But if I am interested only in the question “Am I BB?”, it doesn’t matter how good I am thinking of being a BB or everything else, because the only thing which I am interested in is already assumed as true.
The same logic was used in the attempts to disprove the simulation argument by Danila Medvedev (and maybe others). If I am in a simulation, I can’t have a perfect model of the distribution of simulations in the universe and—Danila suggests, – it disproves the simulation argument.
Let’s elaborate: One of two statements is true:
I am real and simulations are impossible.
I am in simulation, but in that case, I can’t perform perfect probability estimations.
The italic part doesn’t have any logical power over the whole argument.
If you can’t perform any probability estimates, it’s pretty devastating.
If I can’t perform any probability estimates, then I am Boltzmann brain. ))
If you are BB,you can’t perform probability estimates, and Carrol’s argument goes through.
Carroll argument resembles a typical proof by contradiction: let’s assume A, but it produces a contradiction, so A is false. However, in Carroll’s case, no matter what is happening with BB, there is no contradiction, so it is not a proof that I am not BB. This is even stronger in the case when it is applied to the simulation argument, as a sim can think coherently.
But what Carroll actually said is that BB can’t know for sure that it is BB.
But from my point of view, its all is compensated by the dust theory: for any BB1 there is a BB2 which looks like logical continuation of BB1. Like for any random integer N there is another integer N+1.
Therefore, for any BB which sees absurdity, there is another BB which recognise it as absurdity. Thus in dust theory, chains of BBs can calculate probability that they are BB.
It actually is an argument that belief in BBs is self undermining.
“A Bayesian analysis of self-undermining arguments in physics”
http://philsci-archive.pitt.edu/20487/
If you are a BB at all, you have a vanishing likelihood of being a sane BB.
Thanks for the link to Wallace’s work. I think that he also fails in the same trap, that is he fails to distinguish “large” and “small” forms of self-undermining arguments.
The large one is a choice between two alternatives: “1) I am real; 2) I am BB, and therefore can’t have reasonable thoughts.”
The small argument is only (2) from above: “I am BB, and therefore I can’t have coherent thoughts. “Only the small argument is self-undermining but the large one stands because there is no probability update in (1) horn.
For example, I have a 10 per cent chance to believe that I am real a priory, based on theoretical physics. I know that BBs can’t update properly but it is not evidence on which I should update, as it is not news for me.
Another example: I can be either awake or dreaming. During dreaming I can’t effectively decide, am I awake or dreaming and properly calculate the chances of being in a dream. Sometimes I have lucid dreams and sometimes I have very realistic dreams. If we follow Carroll’s logic, we should conclude that dreaming is impossible, because it is cognitively unstable, and it could be that I am dreaming now about commenting in LW. But in real life, we assume that dreaming takes, say, 0.3 time of my subjective experiences and we ignore any probability computations about dreaming that we do during dreaming.