6 ms·
The vicious cycle of reputation systems
- Jarred 16y agoI don't think communities should directly rate the content. With more people the general consensus on what is good and what is bad becomes so subjective, to the point where it's what the user's that just happen to see it end up defining it as either "good" or "bad". Instead, the webserver should simply watch the activity levels of all content, including things like whenever a user hovers over a link and for how long to things like replies. And based on all of the available information it should give it a numerical value defining it's influence on the community. It probably can't be 100% accurate, but it can be pretty close and most likely better than what is in place in most communities.
- osamet67 16y agoThis is along the lines of what we're trying to do. Interesting ideas!
- Jarred 16y agoThanks Another big problem with both publishers and general discussion forums is that the type of content changes at certain times of the day, and more specifically, it changes whenever anything new is published. Why that's a problem is because it means that the community is different at different times of the day. In order to fix that the simple solution is to only enable publishing in intervals. The problem with that though is that people like new and interesting stuff. By giving a lag to new content it will increase the chances of the user's being bored. So, the next step would be only showing content that interests the user based on what's interested them before, and having that content be constantly updated. That just went from a reputation system to an individualized search engine. It's just a thought though.
- osamet67 16y agoYeah, that's more of an internal community thing. What we're trying to do is create an abstraction layer above all taht, that will either be agnostic to or be able to understand these kind of trends. Semantic analysis of content, for example, is one rabbit hole we're not planning to venture into.