My newest (and only!) analytical page is very much an experiment in progress: 

https://see.pessimism.com/analysis.html  

I thought I had figured out that more left- or right-leaning news sites tended to have higher "emotional" scores. When the "see" page was only tracking the Washington Post, Fox, The Wall Street Journal, and the New York Times, the scores on the graph pretty much reflected those expectations. Now that I've included 4 more sites, the scores no longer match my expectations. Time for a new theory, maybe. Or maybe humans are just not as easy to pigeon-hole as we'd like to think. Anyway, I plan to keep working on it. My goal is to automate the analysis of sites against various interesting measures, and automate in such a way that human bias is hard to inject into the process. (And I said "hard," not impossible. Bias is extremely hard to avoid completely.)