It’s hard to say what algorithm would serve you better. Seems like this does what you are seeing it to do. It’s not how I’d do it, but I don’t prioritize unblocked users. To fix this, I’d assign a multiplier for zero blocked users. It might be one so that no blocked users is the same as one mathematically. But maybe free speech is so important to you that you give it a multiplier of 2 or wherever.

active_users / [MAX(1,blocked_users)]

This would be a multiple of 1. Change the 1 to a 0.5 for a multiple of 2.

I didn’t look at the script but it probably has a Max function which just takes the higher of two numbers.