As I remarked in my own essay on the subject (linked by billswift below, keyword "superstimuli"), I don't think the problem comes from the edge cases. I think

As I remarked in my own essay on the subject (linked by billswift below, keyword "superstimuli"), I don't think the problem comes from the edge cases. I think the problem is that if you make a product 5% more addictive than the competition, you can steal 80% of their customers, and this creates an incentive to make things more and more addictive well past the point where it starts to hurt the median customer significantly. The edge cases are just the canaries in the coal mine - the ones who are hurt blatantly, visibly, and undeniably.