Project idea: Take ttyrec and store the terminal stream (multiple even) in an Ogg container. Next sample the stream to reduce the number of "frames" and sync with audio / video streams in the file. And then obviously play it.
How about a NLP extension for your text editor, that shifts words in time, certainty, etc. That is, from "can" to "could", from "went" to "goes" and so on. Also providing synonyms and other constructions in a similar mode to the selection of characters/words in (X11 or Emacs) input methods.