I enjoy Duncan's thinking, and reading this post updated my beliefs on the benefits of using storytelling as an aid in making my points easier to understand to others.

It seems to me like narratives take advantage of the machinery that handles how complex experiences get compressed into tokens that later get referenced as fodder for additional stories. Modern stories, though trial and error, have figured out how to send superstimuli like spikes to this machinery, resulting in tokens far weightier/shinier than anything we encounter in our day to day lives. This is why fictional evidence is such a problem.

This seems very related to Brienne's recent article.