idk why I’m just now seeing your reply … that’s very interesting, and another way that we seem to be thinking along the same lines! The WoT in my own desktop app Pretty Good does the same thing: influence (which is contextual) is the quantity being calculated by the WoT, and the fall off is exponential. There’s a scrollbar to adjust how steep the falloff occurs (a constant between 0 and 1, which is factored in to the weight of each rating). If Alice endorses (not a follow, but similar) Bob in some context, then her influence determines how much weight to give to her endorsement. There are scrollbars so users can adjust the default influence of users about whom no ratings exist. There are multiple other variables to adjust in the control panel, designed to mitigate various sybil attacks, so yes it’s also quite complicated, but not intractably so. Most of the user-adjustable variables can be set to default values and hidden from the user, to be discovered later only if they need adjusting.