So, this also comes up in proposals of informative upvotes. My impression is that the halo effect makes this less effective in practice than it seems like it will be in theory.
What we would like to have is different signals—posts that are informative but wrong have high ‘informative’ scores and negative ‘right’ scores. Or posts that are ‘right’ but not ‘funny’, or ‘funny’ but not ‘right.’ But what I suspect will happen is that the various signals will be correlated together strongly, so that you end up with posts that are informative and right and funny, vs. posts that are uninformative and wrong and boring. At which point you could have just stuck with a single karma rating.
It’s possible that this is thinking about what happens with extreme posts, when what matters is marginal posts—if might be that if you get a single downvote, it’s not one person clicking all three buttons, but instead one person clicking one of those buttons, and so you can figure out which one it is.
So, this also comes up in proposals of informative upvotes. My impression is that the halo effect makes this less effective in practice than it seems like it will be in theory.
Have there been any studies or a thoughtful analysis of the Slashdot’s +5 Funny +5 Insightful system?
Can you go into a little more detail? Readers suffer a halo bias about whom?
What we would like to have is different signals—posts that are informative but wrong have high ‘informative’ scores and negative ‘right’ scores. Or posts that are ‘right’ but not ‘funny’, or ‘funny’ but not ‘right.’ But what I suspect will happen is that the various signals will be correlated together strongly, so that you end up with posts that are informative and right and funny, vs. posts that are uninformative and wrong and boring. At which point you could have just stuck with a single karma rating.
It’s possible that this is thinking about what happens with extreme posts, when what matters is marginal posts—if might be that if you get a single downvote, it’s not one person clicking all three buttons, but instead one person clicking one of those buttons, and so you can figure out which one it is.
Why can’t we have both?