> Amazon can add additional weights based on the user's history and using graph network analysis of that user's review distribution. Bad items don't necessarily need to be banned, but the negative weight can make the listings virtually invisible.
I cannot for the life of me understand why Amazon doesn't already do this.
It's entirely possible that they already are/ have looked at it, and found this move to be more strategically effective.
Ultimately, Amazon's goal is to maintain a strong brand and keep stockholder confidence. This presents a much stronger move without admitting as much weakness, compared to allowing the behavior to continue and trying to mitigate it with AI (since the layperson -- ie most stockholders -- has a poor understanding of the efficacy of ML techniques).
Additionally even if if they did, they are less incentivized to publicize it. Secretly diminishing the voting power of particular reviewers works just as well, if not better, if they don't know it's happening at all, and prevents creation of multiple accounts to avoid said penalties.
I cannot for the life of me understand why Amazon doesn't already do this.