Suppose that my method of assigning probabilities about things that should not update on new evidence of any kind (invisible dragons) is to assign them either a very low probability (something which I almost always do), or a very high probability (something I very rarely do)?
This statement seems insane in that one should always update based only on one’s current evaluation of it is (if you make the change “less than or equal to 30%”, it has a stable solution)
Suppose that my method of assigning probabilities about things that should not update on new evidence of any kind (invisible dragons) is to assign them either a very low probability (something which I almost always do), or a very high probability (something I very rarely do)?
This statement seems insane in that one should always update based only on one’s current evaluation of it is (if you make the change “less than or equal to 30%”, it has a stable solution)