Experience vs. expectations is the only reasonable way to interpret user-based scores even if they are not astroturfed. The problem is that you don't always know what the average user's expectations are and how they compare to your own and there is no basis for comparing scores between different objects.