Less Wrong is a community blog devoted to refining the art of human rationality. Please visit our About page for more information.

Seth_Goldin comments on The Amanda Knox Test: How an Hour on the Internet Beats a Year in the Courtroom - Less Wrong

42 Post author: komponisto 13 December 2009 04:16AM

You are viewing a comment permalink. View the original post to see all comments and the full post content.

Comments (632)

You are viewing a single comment's thread.

Comment author: Seth_Goldin 24 March 2010 03:23:41AM *  3 points [-]

I'm a little late to this game, but I spent over an hour, maybe two, comparing the information from the two websites. I had known nothing previously about the case.

My answers: 1: 0.05; 2: 0.05; 3: 0.95; 4: 0.65

So, I feel pretty vindicated. This was a great complement to Kaj Sotala's post on Bayesianism. With his post in mind, as I was considering this case, I assigned probabilities to the existence of an orgy gone wrong as against one rape and murder from one person. There is strong Bayesian evidence for Guédé's guilt, but it's exceedingly weak for Sollecito and Knox. This has really helped the idea of Bayesianism "click" for me.

komponisto, your reasoning is wonderfully thorough and sound. I can corroborate that I deliberately found myself "shutting the voice out" concerning the activity with the mop. You have a great explanation, overall. These two posts of yours are in the running for my all-time favorites.

Comment author: komponisto 25 March 2010 05:53:30AM *  7 points [-]

Thanks for the kind words!

Actually, looking back, I now think I could have done better. In particular, I wish I had been more explicit about the central probability-theoretic point: the fact that the evidence against Guede screens off Kercher's death as evidence against Knox and Sollecito. This point was missed by a number of commenters; if you read the discussion you'll find various people saying that the prior probability "should" take into account the fact that a murder occurred in Knox's house. In actuality, of course, it doesn't matter where you start, so long as you eventually incorporate all of the relevant information; but what must be understood is that if you start with probability mass assigned to Knox and Sollecito because of Kercher's death, then you have to take (most of) that probability mass away upon learning of the evidence against Guede. In other words, under this setup, evidence of Guede's guilt becomes evidence of Knox's and Sollecito's innocence -- something which is counterintuitive and very easy to forget (with tragic consequences).

This issue of "choosing the prior" and other Bayesian subtleties encountered in these discussions may be worth revisiting at some point.

Comment author: redlizard 17 August 2013 12:32:40AM *  1 point [-]

In particular, I wish I had been more explicit about the central probability-theoretic point: the fact that the evidence against Guede screens off Kercher's death as evidence against Knox and Sollecito.

I think this insight warrants a great amount of emphasis. The fact that Kercher's death is screened off by some factor unrelated to Knox and Sollecito means that the question of whether the given evidence against Knox and Sollecito is sufficient to judge them co-conspirators is equivalent to the question of whether the given evidence against them would have been sufficient to judge them murder-conspirators in the absence of a body. And I don't think anyone believes THAT is the case.