He said, “Well, um, I guess we may have to agree to disagree on this.”
I [Yudkowsky] said: “No, we can’t, actually. There’s a theorem of rationality called Aumann’s Agreement Theorem which shows that no two rationalists can agree to disagree. If two people disagree with each other, at least one of them must be doing something wrong.”
...
Robert Aumann’s Agreement Theorem shows that honest Bayesians cannot agree to disagree
...
Regardless of our various disputes, we [Yudkowsky and Hanson] both agree that Aumann’s Agreement Theorem extends to imply that common knowledge of a factual disagreement shows someone must be irrational.
...
Nobel laureate Robert Aumann—who first proved that Bayesian agents with similar priors cannot agree to disagree
Do you think I’m misunderstanding the sequences or do you disagree with them?
Just because it’s not fully proven in practice by math doesn’t mean it isn’t a broadly true and useful idea.
It is fully proven by the math, but it requires a set of stringent conditions about honesty and shared information which are unlikely to obtain in real world situations. As explained in the rationality article. Did you read it?
It’s not that you misunderstood the summary versions, it’s that the summary versions are inaccurate. In general, you should summarise something as it operates under the prevalent, realistic conditions. So “you can’t use Bayes for everything” and “people aren’t suddenly going to start agreeing, even if they are rational”.
If you read all the way through the rationalwiki article on Aumanns Theorem, there is a clear explanation as to why it cannot apply in practice.
...
...
...
Do you think I’m misunderstanding the sequences or do you disagree with them?
Just because it’s not fully proven in practice by math doesn’t mean it isn’t a broadly true and useful idea.
It is fully proven by the math, but it requires a set of stringent conditions about honesty and shared information which are unlikely to obtain in real world situations. As explained in the rationality article. Did you read it?
It’s not that you misunderstood the summary versions, it’s that the summary versions are inaccurate. In general, you should summarise something as it operates under the prevalent, realistic conditions. So “you can’t use Bayes for everything” and “people aren’t suddenly going to start agreeing, even if they are rational”.