Fair point, what about both the robot and federation creates 1986 events, and the federation one is like a check mark, you either trust one, the other or both.
Another option would be to have some sort of "seal of approval" 🤔 the user rates and the federation just validates, literally would be like a NIP-25 reaction from the federation.
Yes, the federation can add a label to mark authenticated reviews as an additional data point. But a Review platform can do much better by simply understanding the WebOfTrust score of each key sending the review and discarding those keys marked as spam or bots. Think Google's Page Rank for keys.
For instance, if I go to a store, I would prefer to see reviews by the people I am following, or follows of their follows, than any other user even if famous. This of course breaks the single review score calculations. But I would argue it is a more representative to the quality of the review than a simple average of any other filter over the reviews.
