Sorry to go off topic a bit here, but this is something I think about a lot. One of the downsides to personalized news driven by recommender systems is that our biases are constantly confirmed and become ever stronger. This results in more polarization, as far as I can tell, because we're only hearing what's easy to hear. Not only that, but the content providers are more incentivized to deliver news articles that will be directed to their constituents (click counts confirming this). It's not the best recipe for opening minds.
Something I've learned to do when reading all news is to figure out their narrative first, and then read the article in that context. The more quickly you become aware of the narrative they're trying to tell, the easier it is to detect the BS. The most important time to do this is when you're reading something that is confirming your existing biases, as you're much less likely to have your skeptical radar up at that time.
I'm toying with an idea to create a recommender system for news which takes your preferences, and selectively gives you contrarian articles that aren't off-putting, but give you a different viewpoint. The problem is finding content that presents the evidence/news in a less narrative-based fashion.
/end off topic rant