> For the complete documentation index, see [llms.txt](https://distsn.gitbook.io/recommendation-fairness/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://distsn.gitbook.io/recommendation-fairness/designing-fair-user-recommendation-system.md).

# Designing a fair user recommendation system

In this chapter, we design a *fair* user recommendation system for Mastodon, in step-by-step style.

First of all, we have to avoid the oldies effect. We can choose weekly, monthly, or other criteria with some finite duration.

Second, to avoid the shrink-wrap effect, we have better to use the number of boosts, rather than the numer of followers.

Now we find that, the weekly number of boosts is a *fair* criteria to recommend the users.

And one more thing, a reverse *unfair* criteria is *fair*. For example, we can imagine a rule that the user who has more than 100 followers is not recommended. This rule can be combinated with the weekly number of boosts criteria, which brings some additional *fairness*.

Now we get a design of a *fair* user recommendation system for Mastodon. Implementation is expected.