Sure, we all want good recommendations, but how do we define "good"? I see some obvious problems with the way tech recommendations are playing out that have no

Sure, we all want good recommendations, but how do we define "good"? I see some obvious problems with the way tech recommendations are playing out that have nothing to do with the filter bubble issue. First, we are outsourcing recommendations to profit-driven corporations. At best, the recommendations I want to see will be mixed with the ones advertisers want me to see. Revenue expectations of social media companies currently ensure this is a non-trivial issue with no easy middle ground. Even assuming the recommendations are reasonably good, you still have the problem that these ML-type recommendations require massive datasets and therefore it naturally gravitates towards a winner-take-all situation. Having one giant borg of a recommendation engine is bad. We are replacing a rich tapestry of individual word-of-mouth and small-scale communication channels with a massive mono-culture. Scalability is worshipped in tech companies (it's great for getting rich!), but it's not a inherently a good thing. In nature we find diversity trumps scalability, and I feel the same way about culture.