In this gist https://gist.github.com/mayhem/3e344db9e887c88056dce00d78b440f5  we see that the similar users are either users who:

1. have very few listens
2. are spammers

We should pass in a constant from the lemmy front end for the number of listens that a user must have to be included in the similar users matrix. This should clean up the first category of users and then make the spammers rise to the top of the list. Then we can add a report user function so that we can block spam accounts.

This ends up being a great tool since the community will have  motivation to keep the list clean so that similar users scores will work correctly.