As a technologist I couldn't be more impressed by Netflix's rise over the last decade, but the old film artist in me kind of despises the direction they are taking motion pictures.

I recently stopped using it and it was like walking out of an incredibly comfortable, yet subtly depressing cave. I found it hard to get out of the cycle of watching only films that I was likely to watch.

What humans will eventually discover after we've squeezed every last drop of value out of the monolith ML systems of today is that many facets of our culture do not benefit from traditional recommendations systems. Maybe a more democratized and transparent system for model consumption could help. If indie researchers and hackers can integrate their ML models into existing systems and offer truly diverse ML models then maybe recommendations systems have chance. But in their current state they are laughable representations of the average taste of its users, which is generally not great.