I've heard that the outage was a screw-up by a junior programmer that was accidentally given too much access to the system by his/her supervisor. Apparently a bad configuration was applied to their servers. This seems to match up with the official Facebook announcement linked in that article.