The model gets interesting if reviewer credibility is scored by calibration over time, not follower count. Without that, it will drift toward reciprocal endorsements and people gaming each other’s profiles.