I wonder if there has been work on Bayesian spam filtering but for upworthy-style clickbait headlines. Phrase substitution is fun, but I'd love to see some sort of machine learning that could pick up on the clickbaityness of headlines.
For example, I can't remember the last time a headline with "This/These" or "You" wasn't clickbait.
Instead of 'fixing' the hyperbolic headlines, stop following the sources that have shared them, and ignore these sites. Upworthy et. al. do this because it works.
The author's idea was to block upworthy headlines, but I use it myself to block any article links to these upworthy websites, on sites like HN and Reddit.
For example, I can't remember the last time a headline with "This/These" or "You" wasn't clickbait.