= 27d567bc2d48d9f40e9ccc20ea370b15)
Eliezer_Yudkowsky comments on The Amanda Knox Test: How an Hour on the Internet Beats a Year in the Courtroom - Less Wrong
You are viewing a comment permalink. View the original post to see all comments and the full post content.
= 783df68a0f980790206b9ea87794c5b6)
Loading…= b715d02e85f7b3b4ae65ca3d3e450868)
Subscribe to RSS Feed
Powered by Reddit
= f037147d6e6c911a85753b9abdedda8d)
You are viewing a comment permalink. View the original post to see all comments and the full post content.
Comments (632)
A bizarre or unexpected condition of the crime scene is not the explanandum here; Meredith's death is. One person is entirely sufficient to have killed Meredith, and the DNA evidence establishes with virtual certainty that Guede had the kind of contact with her necessary to accomplish this. Unless the evidence suggesting someone else was involved is of comparable power to the DNA evidence against Guede (something on the order of 30 bits), then (and this is the part people have trouble understanding) even paying attention to it at all is automatically hypothesis-privileging.
(EDIT: Eliezer corrects below. What I actually wanted to argue here was that, given the certainty of Guede's involvement, the lack of connection between him and Knox or Sollecito is strong evidence against their involvement -- probably enough on its own to outweigh the comparatively weak evidence against them provided by the alleged indications of multiple attackers at the crime scene.)
Yes, we might be curious about the unusual mechanics of the crime scene given only one person, but unless they are so strange that assuming someone else's guilt of murder (when we already have a suspect) would constitute a reasonable explanation for them, we have to regard the whole question as a distraction.
Doesn't follow. You can have a lot of evidence for one true statement and then less evidence for another true statement.
Not sure I understand the objection. The point is that the evidence is extremely weak by comparison; not strong enough to justify any attention.
Whether evidence is strong enough to justify attention is an absolute threshold. There is no "extremely weak by comparison". There is just "extremely weak".
It is extremely weak on its own, and its weakness is compounded and confirmed by the strength of evidence of someone else. The reason for this is that the strong evidence sets up a perameter, a reference point of what is possible for evidence left behind. It puts lines on the thermeter by which to read the murcury. This is because there is also no evidence of collusion, so the physical evidence has to carry most of the weight, if not all. Otherwise, the prosecution's case operates via a tautology.
Huh? Attention isn't binary, off-or-on; like evidence itself, it's a quantifiable commodity. The stronger the evidence, the more attention. Right?
There will be some threshold of evidence below which a hypothesis ought to receive strictly zero attention. You could probably even formalize this in terms of bounded rationality.
Right, but I don't need to claim that the anti-Knox evidence is below that threshold. Unless, that is, we're talking about extremely imperfect less-than-Bayesian human minds, who can't intuitively perceive the difference in weight that a perfect Bayesian would assign to 30-bit evidence vs. 10-bit evidence.