I know the thread is a little old, but I found a link to a blog posting yesterday that made me think of this thread. I figured I'd post it for those who might be interested:
http://www.michaeleisen.org/blog/?p=358 . It outlines in detail basically what kirk has already mentioned.
The long and short of it -- 2 large sellers use algorithmic pricing to set their price relative to the other's price. If the algorithms interact in a funny way it can spiral out of control (example in the blog -- Seller B sets their price to .998 * Seller A's price, Seller A resets their price to 1.27*Seller B)