Thinking backwards

An analogy can't establish truth, only provide intuitive understanding

  • truth must be established otherwise. a brilliant analogy proves nothing.
  • and here's why, I believe. it's all about the Bayes. so what Bayes's Theorem does is tell you how likely something is, given a prior and some new info. so just looking at the new info, it may sound plausible, and so the whole thing can be true. unless, that is, the prior is unlikely. so we tend to go backwards, from a reasonable-sounding conclusion, to an unlikely prior.
  • if B is true given A, that's not evidence that A is true
  • examples:
  • The Amanda Knox Test: How an Hour on the Internet Beats a Year in the Courtroom - Less Wrong -
  • the Amanda Knox case is based on her acting suspicious, and so if the prior were likely (that she had killed), the killing would be a good explanation for the "suspicious" behavior. but if we put aside that behavior, the prior likelihood is very low, so the suspicious behavior counts hardly at all.
  • The Amanda Knox Test: How an Hour on the Internet Beats a Year in the Courtroom - Less Wrong -
  • any instance where a person's character is invoked to explain their (political or other) opinions
  • pretty much everything Rush Limbaugh has ever said
  • re peak oil: "of course liberals believe oil is running out. they're pessimistic about human nature, and anti-freedom." if we were pretty sure we were not running out of oil, it would make sense to try and find out why some people believe differently. but we can't then use a putative reason for such beliefs to go back and use that to try to prove the prior (that we're not running out of oil)
  • it's not that the posterior has no bearing on the prior, but rather that we can't go backwards. it works both ways. If I can use psychological or character-type findings to try and prove facts, I can prove anything. if liberals being over-pessimistic proves we have enough oil, then conservatives' over-optimism proves we don't.
  • bringing it back to analogies:
  • I say opposition to same-sex marriage is wrong, just as opposition to mixed-race marriage is wrong
  • my interlocutor says "but they're not the same"
  • and of course, they're right about the two things not being the same, but they're wrong because they're wilfully missing the point
  • but it's my fault in the first place for using an analogy
  • if opposition to same-sex marriage were indeed wrong, I should show how by pointing to the wrongness of opposition to mixed-race marriage.
  • or I could rightly point to mixed-race marriage as a way that same sex marriage could be okay
  • but I can't go backwards. I can't say that because mixed-race marriage is okay, so is same-sex marriage.

Is this related to Bayesianism?

  • We often look at an effect and try to deduce the cause, but this might be misleading by itself. We should also look at the putative cause and see how often it could lead to the effect of interest.thinking backwards….rather than start at the beginning, start at the ends and see which of them could have resulted from the beginning conditions
  • lots of stuff like this on the mock GRE I took
  • examples I've found since then:
  • Three logicians walk into a bar. The bartender asks "Do all of you want a drink?" The first logician says "I don't know." The second logician says "I don't know." The third logician says "Yes!"
  • that thing on LessWrong about the American woman who was accused of murdering her Italian boyfriend
Unless otherwise stated, the content of this page is licensed under Creative Commons Attribution-ShareAlike 3.0 License