4 ms·

If you dismiss all the evidence "supporting this theory" (your words) then obviously you wouldn't find anything that supports it. Evidence does not have to be

by hnfong 3y ago

If you dismiss all the evidence "supporting this theory" (your words) then obviously you wouldn't find anything that supports it. Evidence does not have to be perfect. Even biased evidence should update your priors. There is evidence supporting this theory. It doesn't mean it is definitely true.

gruez 3y ago
> If you dismiss all the evidence "supporting this theory" (your words) then obviously you wouldn't find anything that supports it. Right, and the converse is you accepting any evidence that vaguely conforms to the narrative of "he got killed by boeing", regardless of how shaky it is. >Evidence does not have to be perfect. Even biased evidence should update your priors. There is evidence supporting this theory. It doesn't mean it is definitely true. This feels like a motte and bailey. You claim that we should update our priors even based on biased evidence, but I never claimed otherwise. My previous comment merely pointed out the issues with taking statements like "if anything happens don't assume it's a suicide" at face value. Moreover the comment I was replying to wasn't merely claiming that we should update our priors, it was that the evidence in question was "plenty, together with the obvious motive, to form an opinion". That's a much stronger claim.
- hnfong 3y ago
  > Right, and the converse is you accepting any evidence that I literally said one must update their priors, not just "accepting" it. Or maybe updating priors is a form of "acceptance", but now you claim it isn't a problem: > You claim that we should update our priors even based on biased evidence, but I never claimed otherwise The comment you replied to just answered your question: "is there anything supporting this theory". Presumably everyone here knows basic Bayesian probability, or at least knows how to weigh supporting evidence without just blindly accepting it. Claiming otherwise seems to be rather uncharitable interpretation of the comment you replied to.
  - renewiltord 3y ago
    Watching people with binary-valued probability measures talk to people with multivalued probability will never not be amusing. The resulting conversations have such fundamental disagreements.

by hnfong 3y ago

gruez 3y ago

> If you dismiss all the evidence "supporting this theory" (your words) then obviously you wouldn't find anything that supports it. Right, and the converse is you accepting any evidence that vaguely conforms to the narrative of "he got killed by boeing", regardless of how shaky it is. >Evidence does not have to be perfect. Even biased evidence should update your priors. There is evidence supporting this theory. It doesn't mean it is definitely true. This feels like a motte and bailey. You claim that we should update our priors even based on biased evidence, but I never claimed otherwise. My previous comment merely pointed out the issues with taking statements like "if anything happens don't assume it's a suicide" at face value. Moreover the comment I was replying to wasn't merely claiming that we should update our priors, it was that the evidence in question was "plenty, together with the obvious motive, to form an opinion". That's a much stronger claim.

hnfong 3y ago
> Right, and the converse is you accepting any evidence that I literally said one must update their priors, not just "accepting" it. Or maybe updating priors is a form of "acceptance", but now you claim it isn't a problem: > You claim that we should update our priors even based on biased evidence, but I never claimed otherwise The comment you replied to just answered your question: "is there anything supporting this theory". Presumably everyone here knows basic Bayesian probability, or at least knows how to weigh supporting evidence without just blindly accepting it. Claiming otherwise seems to be rather uncharitable interpretation of the comment you replied to.
- renewiltord 3y ago
  Watching people with binary-valued probability measures talk to people with multivalued probability will never not be amusing. The resulting conversations have such fundamental disagreements.