Thanks for the pointer. John Nerst's approach is similar to mine.
The way I would formulate it here : De facto, people have different priors. If there is a debate/discussion, the most fruitful result would come by construing, in common if possible, a more encompassing reference frame, where both sets of priors can be expressed to their respective satisfaction. It is not easy. Some priors will be incompatible as such. A real dialogue supposes a readiness to examine ones priors and eventually adjust them to be less restrictive. A static defense of one's priors is mostly a waste of time (or a show).
Caveat : bad faith exists, people, and groups, have vulnerabilities they will protect. So a real dialogue is not always possible, or only very partially.
Thanks for the link. It is quite good. A limitation, that the author would not have been aware of when starting, but is now "glaring", is that his AI's are "rule based".
Thanks for the pointer. John Nerst's approach is similar to mine.
The way I would formulate it here :
De facto, people have different priors.
If there is a debate/discussion, the most fruitful result would come by construing, in common if possible, a more encompassing reference frame, where both sets of priors can be expressed to their respective satisfaction.
It is not easy. Some priors will be incompatible as such.
A real dialogue supposes a readiness to examine ones priors and eventually adjust them to be less restrictive.
A static defense of one's priors is mostly a waste of time (or a show).
Caveat : bad faith exists, people, and groups, have vulnerabilities they will protect. So a real dialogue is not always possible, or only very partially.
The idea is to at least try.