This is why in my system I have more specific, falsifiable metrics: freshness, confidence, etc. which come together to create a fitness score at the surface-lev

This is why in my system I have more specific, falsifiable metrics: freshness, confidence, etc. which come together to create a fitness score at the surface-level, while still exposing individual metrics in the API.