[dang]

Good point, and it looks like something that can easily be scraped out. I'll add it to the list. Thanks!

[Semiapies]

Maybe we should use some machine learning solution that recognizes phrases that look too much like titles.

[dang]

The HN title dataset is public. If anyone can come up with software (ML or otherwise) to detect baity titles and/or debait them, we'd certainly be interested.