RSS doesn't need more machine learning. It doesn't need "algorithms". It doesn't need a business model. It doesn't need to be friendly towards tracking and monitoring user behavior. It absolutely shouldn't care about branding.
Really RSS should just be about reading, and nothing else. The client can do fancy processing of feeds, but the protocol doesn't need to care about that.
@bob that's kinda how Newsblur handles it. It takes the RSS feeds and then has an "intelligence trainer" that you can use to modify the feeds. It works for me.