Let’s say we have statement X that we ascribe a prior probability of 0.3
Then let’s say we have a news source N, presented as an infinite feed of statements. N is sensationalistic with regards of X that are interesting to N:
If X is true N virtually always says that X is true.
But if X is not true N might still say that it is true.
N only reports if it intends to say that X is true, otherwise it says nothing and covers other topics instead.
The probability of N saying X is true when X is not true for our particular X is not known, but historically we know that N is wrong about 70% of times.
X is interesting to N and N says X is true. How will this shift our belief in X?
[Question] Bayesian update from sensationalistic sources
Let’s say we have statement X that we ascribe a prior probability of 0.3
Then let’s say we have a news source N, presented as an infinite feed of statements.
N is sensationalistic with regards of X that are interesting to N:
If X is true N virtually always says that X is true.
But if X is not true N might still say that it is true.
N only reports if it intends to say that X is true, otherwise it says nothing and covers other topics instead.
The probability of N saying X is true when X is not true for our particular X is not known, but historically we know that N is wrong about 70% of times.
X is interesting to N and N says X is true.
How will this shift our belief in X?