-
@ david
2025-02-14 20:11:12
And an addendum: what’s the theoretical reason for the inescapability of incorporating nonlinearity into centrality algorithms? I’d argue it’s because our CONFIDENCE in a calculated score (like the average of 1-5-star ratings) is a fundamentally important term in calculating trust weights. CONFIDENCE is basically calculated by mapping a number from 0 to infinity (a measure of the available evidence that we use in calculating a score, e.g. the number of ratings) onto the interval 0 to 1, and a mapping from [0, infinity) to [0,1] must of necessity be nonlinear.