There are a couple of cases where the structure of the page means my XPath query is wrong but in the vast majority of cases, the XPath is correct.
I assume the reason is due to the error message of Max regex input length of 17500 exceeded however I don't know a way to fix this. I'm using the same format for a lot of other pipes with no problems at all, it's just Gizmodo and Lifehacker that are causing these issues.
I've tried removing all of the regex expressions apart from the first one (which replaces the description text) to no avail.
Hmmm, not sure I did fix the error as I just reran the cloned pipe that I had edited and I still got it. That explains why I was still getting blank articles.
I guess this is just a quirk of the Gawker sites and is a bit frustrating as they've only just turned-off their full-text feed. I did find a great service at fullrss.net which uses a centralised database of XPath information for sites. This seems to be working perfectly for Gizmodo and Lifehacker so I'll stick with that.