> Look for emotional language in articles. Then count up the hard facts. There's an inverse relationship. It's not supposed to be like this! Seems like it woul

> Look for emotional language in articles. Then count up the hard facts. There's an inverse relationship. It's not supposed to be like this! Seems like it would be great to train a model on the historical corpus of news articles to visualize this trend and inform readers of this kind of deliberate manipulation. It might also be interesting to have a browser plugin that could use the model to "rewrite" articles to remove the manipulation as well.