Invisible Ratings

You can criticize the principles, specific formulas by which ratings are calculated on Habré, but this is not about that. And not even that Habr has no competitors where one could go in the hope of a different rating solution. The problem is wider - building a good rating system is an extremely difficult task, depending on the type of resource, the composition of users, human psychology, and so on. Here are a couple of examples of studies on how ratings can be related to audience composition and type of resources [ 1 ], [ 2 ]. I can’t get around my beloved [ 3 ]. According to the mind, you need to have a whole research department with highly professional psychologists, mathematicians, sociologists, Internet specialists, etc., to build something acceptable. But if you follow the logic of my postGood projects are created by users , and this is not enough.

Suppose that Habr users have the ability to take the results of other people's votes on karma, topics and comments and (somehow) calculate the values ​​of karma and rating of participants, according to their own approaches and formulas. These alternatives, for example, will not be visible to anyone except the rating author. But based on them, you can automatically form your own top materials. The idea is similar to the idea of ​​friends on blog platforms - this tape is essentially our individual top. The difference is that we have to indicate friends there “by hands”, and here they will fall into our stream according to the results of voting by strangers, but according to our rating principles. Of course, this is rather a toy for geeks, for those who are interested.

Probably the most interesting thing to do with such things is the developers of Habr and similar resources - for example, you can make 10 or 100 (as many as you can) invisible alternative rating options and then watch which one is the best (in your opinion, of course) identifies the most interesting authors and materials . This is essentially close to creating 10 or 100 Habr options with the same content and without the need for users to vote anew in each of them.

This approach has a drawback - if you change the rating system explicitly, apparently and without alternatives, then users will slightly change their behavior in response. Those. you have an implicit response. With invisible ratings, this response is not. However, we can read other people's friends. Accordingly, we can conceive of a situation where alternative sortings of content and authors, if desired, will be visible to all users. And you can even vote for these options :) At the same time, it’s possible that not the Habr’s developers, but one of the users will win. Just kidding - it’s not just that I intend to oppose the professionals and the “crowd”, but I only want to emphasize the value of mass experimentation in complex fields.