I’m seriously thinking of building my own Markov chain generator using KJV and tweets tagged #tcot. I wonder what you’d get from this project if you used two-word keys; I did that with some Mark Twain and it got weird, but still readable. Maybe the differences in 500-year-old English and 100-year-old English are too great for it to be effective.
After working on another project not related to Markov chains, I’d point out that if you snag KJV off of Project Gutenberg, all the verses end with a newline and begin with a CHAPTER:VERSE marker. Easily parseable in Perl or (in my case) Ruby.
