but crucially, it assumes that variations in submissions and voting activity are strongly correlated, which I imagine is not necessarily the case