I must disagree with the comment of pmjordan. My impression is that the math argument didn't made through. If the issue is raised over and over their might be

I must disagree with the comment of pmjordan. My impression is that the math argument didn't made through. If the issue is raised over and over their might be a possibility that it has not been optimally addressed yet. I also understand the initial question as an invitation for suggesting for a better solution. Why smashing this with a "don't ask" comment ? This is a hacker topic and it has many potential applications if a better solution is found. Back to the initial topic. It looks like the problem comes from using a simple scalar parameter for the ranking. There are other parameters in play that could be combined to yield a more pertinent ranking: number of leads, number of votes, kind of votes (good, bad, off topic, spam, ... ), number of comments. The ranking could also be multidimensional instead of a simple scalar value. For instance, identify hot topics, popular topics and pertinence, etc. 10 up votes may mean 90% readers up voted the subject or only 1% up voted the subject which is a very different information value. The google story, as well as the reddit story, should remind us how important the ranking can be. The more pertinent your information channel becomes the more valuable it is.