cupholder comments on Undiscriminating Skepticism - Less Wrong

97 Post author: Eliezer_Yudkowsky 14 March 2010 11:23PM

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

Comments (1329)

You are viewing a single comment's thread. Show more comments above.

Comment author: cupholder 23 March 2010 03:52:20PM 1 point [-]

On the contrary, it is quite possible that there could be evidence that would convince me of either of those things. It is just that the evidence would have to be strong enough to go head-to-head with basic physics. If it could somehow be demonstrated that Avogadro's number were 300 orders of magnitude too tiny, and that molecules were a googol times smaller than we thought, and could explain why our earlier experiments had led us to our original estimates of Avogadro's number and molecular sizes, then that would tend make the effectiveness of homeopathy (more) plausible.

Comment author: brazil84 23 March 2010 04:21:39PM -1 points [-]

It is just that the evidence would have to be strong enough to go head-to-head with basic physics.

And by what standard would you decide whether the evidence is sufficiently strong?

Comment author: cupholder 23 March 2010 08:58:25PM 0 points [-]

My estimate of the probability of homeopathy working and the current laws of physics being very different would have to be of similar order to my estimate of the probability of the current laws of physics being correct.

Comment author: brazil84 23 March 2010 11:49:15PM *  0 points [-]

And how do you come up with your probability estimates in a situation like this? Do you rely on your general knowledge and common sense? Do you have some algorithm you follow?

Comment author: cupholder 24 March 2010 12:32:03AM -1 points [-]

No, I don't have a strict algorithm I follow in situations like this. What I actually do is probably more like this:

  • do some initial reading to get an idea of the basic plausibility of the hypothesis based on my background knowledge
  • let the hypothesis bounce around my mind for a while
  • try to spell out to myself the resulting gut feeling for the hypothesis' probability
  • check that rough estimate for any gaping flaws
  • if that rough estimate is really low, reject the hypothesis as Too Unlikely To Debate for the time being (remember that 'super careful' warning I made a few posts up? This is where it applies)
  • if the rough estimate is instead very high, accept the hypothesis as Too Likely To Debate for the time being
  • if the probability estimate is more middling, and the hypothesis' truthiness is important to me, gather more data and try to hone my hunch for the hypothesis' probability