How does NewsBlur know whether I like or dislike a story?
When you like or dislike a story, you mark a facet of that story by checking a tag, author, part of the title, or entire publisher. When these facets are found in future stories, the stories are then weighted with your preferences. It is a very simple, explicit process where you tell NewsBlur what you like and don't like.
The idea is that by explicitly telling NewsBlur what your story preferences are, there is increased likelihood that you will like what the intelligence slider does for you.
Currently, there is not an automated way of detecting stories you like or dislike without having to train NewsBlur. This implicit, automatic intelligence will come in the near-term future, but it will require an evolution to the interface that has not been easy to figure out how to make in a simple, clear, and effective manner. Soon.