I love this paper on communication styles in the Linux kernel:
Training a classifier to distinguish two particular authors, leaders in the Linux
development community, based on lexical choices. Use of "sorry",
"thanks", "actually", "never" and expletives are most
It makes me wonder whether this would also be an interesting characteristic of one mailing
list compared to another, in addition to distinguish individual authors. "Where does
your open source community fall on the Actually-Thanks Spectrum (TM)?"
I'd love to see this as part of BigBang, particularly if that kind of lexical analysis
or Bayes classification would be useful for lots of research questions.