October 23rd, 2007


I spent an interesting evening with a grain of salt

What happens when you feed my 20 most recent LJ entries through a Markov chain?

Nothing, actually.

What happens when you feed my 40 most recent LJ entries through a Markov chain?

Well, still nothing, actually.

But what happens when you take my most recent 40 LJ entries, strip out all the timestamps and "Leave a comment" and "Current Mood"s, and then tack on a few handpicked long entries on a variety of subjects (specificially improv theater, video games, books, religion, and my love life), representative of the range of my writing style, and feed that through a Markov chain?

You get...a long wait!

But after that you get comedy gold.

Collapse )

Aside from its inability to do anything interesting with less than a couple thousand words of text, this algorithm totally trounces all those random LJ generator toys out there, even though it's considerably less sophisticated than some of them. Mark V. Shaney's revisions sound a little bit like what goes through my head right before I fall asleep, with my true self already zoned out but my language synapses still firing.

If any of you guys are willing to spend the time and effort to make a long sequence of just entry text (no moods or current musics or times and dates) and feed it through the script, I'm curious to see what you get.

For more on Mark V. Shaney, click here.