Still, if you do in fact regard Yudkowsky’s (or anyone else’s) stated belief in some proposition as being completely uncorrelated with the truth of that proposition, then you should not revise your belief in either direction when you encounter that statement of his. Otherwise, failure to update on evidence – things that are entangled with what you want to know about – is irrational.
This is accurate in the context you are considering—it is less accurate generally without an additional caveat.
You should also not revise your belief if you have already counted the evidence on which the expert’s beliefs depend.
(Except, perhaps, for a very slight nudge as you become more confident that your own treatment of the evidence did not contain an error, but my intuition says this is probably well below the noise level.)
You should also not revise your belief if you have already counted the evidence on which the expert’s beliefs depend.
What if Yudkowsky says that he believes X with probability 90%, and you only believe it with probability 40% - even though you don’t believe that he has access to any more evidence than you do?
Also, I agree that in practice there is a “noise level”—i.e. very weak evidence isn’t worth paying attention to—but in theory (which was the context of the discussion) I don’t believe there is.
What if Yudkowsky says that he believes X with probability 90%, and you only believe it with probability 40% - even though you don’t believe that he has access to any more evidence than you do?
Then perhaps I have evidence that he does not, or perhaps our priors differ, or perhaps I have made a mistake, or perhaps he has made a mistake, or perhaps both. Ideally, I could talk to him and we could work to figure out which of us is wrong and by how much. Otherwise, I could consider the likelihood of the various possibilities and try to update accordingly.
For case 1, I should not be updating; I already have his evidence, and my result should be more accurate.
For case 2, I believe I should not be updating, though if someone disagrees we can delve deeper.
For cases 3 and 5, I should be updating. Preferably by finding my mistake, but I can probably approximate this by doing a normal update if I am in a hurry.
For case 4, I should not be updating.
Also, I agree that in practice there is a “noise level”—i.e. very weak evidence isn’t worth paying attention to—but in theory (which was the context of the discussion) I don’t believe there is.
In theory, there isn’t. My caveat regarding noise was directed to anyone intending to apply my parenthetical note to practice—when we consider a very small effect in the first place, we are likely to over-weight it.
What you are saying is that insofar as we know all of the evidence that has informed some authority’s belief in some proposition, his making a statement of that belief does not provide additional evidence. I agree with that, assuming we are ignoring tiny probabilities that are below a realistic “noise level”.
As you said this is not particularly relevant to the case of someone appealing to an authority during an argument, because their interlocutor is unlikely to know what evidence this authority possesses in the large majority of cases. But it is a good objection in general.
This is accurate in the context you are considering—it is less accurate generally without an additional caveat.
You should also not revise your belief if you have already counted the evidence on which the expert’s beliefs depend.
(Except, perhaps, for a very slight nudge as you become more confident that your own treatment of the evidence did not contain an error, but my intuition says this is probably well below the noise level.)
What if Yudkowsky says that he believes X with probability 90%, and you only believe it with probability 40% - even though you don’t believe that he has access to any more evidence than you do?
Also, I agree that in practice there is a “noise level”—i.e. very weak evidence isn’t worth paying attention to—but in theory (which was the context of the discussion) I don’t believe there is.
Then perhaps I have evidence that he does not, or perhaps our priors differ, or perhaps I have made a mistake, or perhaps he has made a mistake, or perhaps both. Ideally, I could talk to him and we could work to figure out which of us is wrong and by how much. Otherwise, I could consider the likelihood of the various possibilities and try to update accordingly.
For case 1, I should not be updating; I already have his evidence, and my result should be more accurate.
For case 2, I believe I should not be updating, though if someone disagrees we can delve deeper.
For cases 3 and 5, I should be updating. Preferably by finding my mistake, but I can probably approximate this by doing a normal update if I am in a hurry.
For case 4, I should not be updating.
In theory, there isn’t. My caveat regarding noise was directed to anyone intending to apply my parenthetical note to practice—when we consider a very small effect in the first place, we are likely to over-weight it.
I think we are basically in agreement.
What you are saying is that insofar as we know all of the evidence that has informed some authority’s belief in some proposition, his making a statement of that belief does not provide additional evidence. I agree with that, assuming we are ignoring tiny probabilities that are below a realistic “noise level”.
As you said this is not particularly relevant to the case of someone appealing to an authority during an argument, because their interlocutor is unlikely to know what evidence this authority possesses in the large majority of cases. But it is a good objection in general.