Chuck picked out ideas, concepts or just categories of things that he found interesting. The stories then cluster around those ideas. It's interesting how these clusters throw your perception of each story, shifting them from their normalcy.We discussed the issue with keywords and their banality. They don't really let you in. However these 'ideas' are interesting, perhaps because they are more than one word. Compare also to textarc and the ‘association list’. Here the program does not search for individual words, but rather what words are grouped next to each other “mock-turtle” “king-queen” etc. This seemed a very promising way to understand what keywords might be.

An alternative approach would be to place keywords according to a 'thesaurus', so that at the bottom of each story you would have a semantic relationship of the keywords used.