| One thing I wish was different about the way Reddit's algo worked is the way it treats downvotes. Basically, if I am average Joe redditor, there are two potential reasons for a downvote: 1) The post is something I personally disagree with. 2) The post was written by a moron (for various definitions of the word moron). What I wish is that someone (smarter than myself) could come up with a way to reliably determine the motive behind a downvote. I know that's no simple task. But I actually want posts that are slightly more controversial to rank higher than posts which don't generate a lot of downvotes, simply because I'm more interested in reading something thought-provoking than checking out the latest batch of meme-gen. Basically I want comment rankings to look like this: a) Lots of downvotes, few upvotes == total asshole. Maybe racist, maybe a general troll, but I want this ranked at the rock bottom. b) An equal number of upvotes vs. downvotes == flamebait. I'm not interested in the latest holy wars, so we call this the next lowest in rankings. c) Lots of upvotes, few downvotes == uninteresting. Rank it lower. d) The ratio upvotes/downvotes is greater than 1, but less than some finite number N. These are the stories I want to read. I want a ranking system that works something like the above, albeit with more finesse (e.g. perhaps we swap the ranking position of b and c). I'm just not interested in the latest round of adorable kitten pictures, and on HN, too much emphasis on the sheer number of upvotes manifests as groupthink. Also, I want the downvotes that are received by slightly controversial stories/comments (again, slightly controversial as opposed to PC-vs-Mac style flamebait) to have a smaller effect on the poster's karma, so that people will feel free to express disagreement on sites like reddit and HN. |
But why do you downvote controversial stuff then? Why not give upvote for interesting postings and downvote for boring postings?