What do you get when you combine the basic idea of a Markov chainer with the contextual searching abilities of Google? Why, Google Talk, of course. (Thanks to Nyperold for the link…)

Here’s just a sample of the insanity that results from chaining such a large database of text:

“It was a Dark and stormy Night. The rain drenched not only my own thoughts and feelings from the heart. a Leader S Guide. to the Internet. and the World Wide Web Resources. for the study of The American Medical Association. All rights reserved…”

Edited to add: Even better, mentioning a particular anime convention causes us to get into a loop that goes on forever:

“Anime Weekend Atlanta By car from This site. is not a Geek/ ForumOC, Members: Threads: Posts: Last post: The latest version of the United States House of REPRESENTATIVES. Category: of One: s Own ORGANICS Guide to HTML? and CSS. by Example By Example, By Example, By Example, By Example, By Example, By Example, By Example, By Example, By Example…”

And one last bit of insanity, via a mention of that popular cartoon villain:

“Strong Bad Email Addresses. and phone Numbers. and addresses, Phone, Numbers and addresses, Exceptional circumstances EC) drought assistance, for the web, Yellow pages directory ON the internet. This Is London is the sky blue Sounds. like A Pirate Day! Sept. th Sept. th Sept. th Sept. th Sept. th Oct. th am Dr. Megavolt; and his OWN words. A M”

Hilarious stuff, at least for people as easily amused as myself. And it appears to give different results on each run, or at least so it seems, judging from the results I’ve seen on consecutive runs of the same phrase.