I read a lot of technical articles containing website addresses an IPs.
If I use the (official) Instapaper bookmarklet to save an open Newsblur story to Instapaper it creates a seperate article for every IP address or URL mentioned in the article. So, if I read an article that contains 10 IP adresses I end up with 11 Instapaper articles - the one I wanted to bookmark and 10 garbage ones for the IPs.
I confirmed with Instapaper that the behaviour is not triggered by them.