A few days ago I added a regex to the code which generates Hacker Newspaper:
Unfortunately, a YC job ad made it past this filter almost instantly:
(If you're wondering why the comments page looks different: omgwtfhax.)
I felt tricked. I also felt the tactic was a little sleazy and desperate. And the whole reason I added this regex to my HN filter in the first place was to screen out a job ad for Curebit, a company which is best known for seeming a little sleazy and desperate as well.
YC job ad posts, and similar posts on Hacker News, appear with no comment link. This privilege makes an easy target. It should be possible to filter out such content based on that characteristic. However, it's much easier to simply filter out any story where the comments URL is the same as the story URL, i.e., any story which links back to Hacker News itself.
This also excludes "Ask HN" and "Tell HN" stories, along with a few random polls. Not long ago, I would have called that a side benefit, but today I'm not as stridently disdainful of the Hacker News community as I used to be. Instead, I'm going to call it a very acceptable cost for the convenience of not having to spend serious time hacking anything complicated.
However, if you've got the time, the project's on GitHub, and patches are welcome. Also, while I'm at it, I'm very interested in hacking in automated comment elitism -- basically, a way to show comment links when (and only when) comments appear from people I consider more interesting than average. It's a bit obnoxious, but I'm a busy guy, and since discussions wander, I only want to see links to discussions which have a strong likelihood of not wasting my time.