Paul, over the years we all discussed a lot of times about the flaws we find using this kind of naive evaluation system. Now to get rid of some of the problems

Paul, over the years we all discussed a lot of times about the flaws we find using this kind of naive evaluation system. Now to get rid of some of the problems we obv. try to optimise the flawed model or the methods we apply to it. But didn't we learn here on HN that if your model is flawed in a way you can't accept, you should find another model? Just an idea: Maybe a recommendation system would perform better then a voting system. Because we want to sort our posts and comments according to relevance to the reader. And of course there are articles, which are better then others in an object, global fashion. But most of the time 2 articles or comments of the same comment can have quite a big difference in relevance to 2 different people. That is just my simple, unvalidated idea. It maybe totally wrong. I just use it as an argument to the idea that we should not discuss about things like "show comments yes/no" but "how do we better show a reader the stuff that is relevant to him?"