Frequently Asked Questions
- How to format your post.
We support a subset of HTML and you can also format your posts with the help of Textile. Image embeding and some popular video embedding codes (like youtube.com) are also allowed.
- How is it different.
You can tag any post or comment. You can negate tags by preceding them with the symbol '!' (without the quotes).- How it works.
The following is an experiment in online reputation systems. Several attempts at building robust online reputation systems has been made.
Zizindrin.com is our take on this. Ours is implemented as a positional reputation system1.
We will model a discussion forum as a dynamical system of interacting agents. Agents are described by vectors in a (in principle) infinite dimensional space,
where each axis (base vector) corresponds to a finite string of symbols or characters we will call tags.
v = v
1
*tag
1
+v
2
*tag
2
+..v
n
*tag
n
Tags are 'a' 'aa' 'aaa' .. 'b' .. 'bad' .. 'troll' .. 'funny' .. 'z' .. and so on ad infinitum; v
i
belongs to S
1
the unit circle
In our forum users will evaluate each other's actions by tagging them with words (tags). Every time a user tags another user's post each one's position will change relatively in the following way:
Lets say user A tags a post by user B with the word 'interesting', what will end up happening is that the user vectors will change in such a way that their final positions will be closer along the 'interesting' axis, they will attract each other. There are also negative tags, when applying a negative tag user's vectors will move away from each other. The main idea is that users with similar tastes will clusterize. There is also a way of negating tags by simply suffixing the symbol !, i.e. '!good'.
In this model each user's view content will be rated or ordered according to the distance between the reader and the poster. Each user will have a local view of the social network.
(1)
M Voss, L Brückner, A Schlosser -
Positional Reputation Systems
- How is the entropy calculated.
The entropy of a post is the sum of (log p(i,j))/p(i,j) for every descendant comment Ci,j where i is the parent poster and j its commenter andp(i,j) = #(posts where j answers i)/(#total number of posts of i)