I pretty strongly disagree with your takes in that Twitter thread (though agree on the object level that you offered <$1000 too). I think the core point of “we want to find the truth here, and creating incentives against finding the truth seems not worth it if it could damage truth seeking” seems obviously reasonable. This incentive exists even at smaller amounts of money.
I don’t think things like pre-registration are good solutions to this—a clinical trial can maybe register clear rules to follow and a success criteria, in a way that “try to do research + literature reviews on a hard question and understand it better” can’t (Also, clinical trials obviously have major biases due to publication incentives which makes their research much lower quality).
In order for not taking the bet to be a meaningful downwards update, you need to be able to fully decouple “making an epistemic bet given your best guess about reality” from all of the second-order effects of how the bet affects other relevant things. I think the response “I am not capable of this decoupling, and so refusing this bet is not a strong statement about my epistemic beliefs or confidence in them” is extremely reasonable.
He’s also significantly more combative in that thread than I would expect in the context, which leads me to wonder whether there’s more going on there.
Going through it again, I think I was about as combative as you’d expect someone who strongly disagrees with the object-level theory to be. Maybe I could have worded things more nicely. But seriously, there’s not something more going on here.
I am OK betting smaller amounts. I said “up to” $1000.
I pretty strongly disagree with your takes in that Twitter thread (though agree on the object level that you offered <$1000 too). I think the core point of “we want to find the truth here, and creating incentives against finding the truth seems not worth it if it could damage truth seeking” seems obviously reasonable. This incentive exists even at smaller amounts of money.
I don’t think things like pre-registration are good solutions to this—a clinical trial can maybe register clear rules to follow and a success criteria, in a way that “try to do research + literature reviews on a hard question and understand it better” can’t (Also, clinical trials obviously have major biases due to publication incentives which makes their research much lower quality).
In order for not taking the bet to be a meaningful downwards update, you need to be able to fully decouple “making an epistemic bet given your best guess about reality” from all of the second-order effects of how the bet affects other relevant things. I think the response “I am not capable of this decoupling, and so refusing this bet is not a strong statement about my epistemic beliefs or confidence in them” is extremely reasonable.
He’s also significantly more combative in that thread than I would expect in the context, which leads me to wonder whether there’s more going on there.
Going through it again, I think I was about as combative as you’d expect someone who strongly disagrees with the object-level theory to be. Maybe I could have worded things more nicely. But seriously, there’s not something more going on here.
I don’t really expect people to be combative at all in this context, but I’ll happily take you at your word and retract the comment.