Honestly, the real problem is that ragebait works. Like, if that's what people decide to engage with on social media, then that's what's going to show up, whether it's skimmed highly-scoring Reddit content or not.

You could try to produce "no-ragebait" communities.

You could try to introduce a recommendation algorithm that uses heuristics to try to identify ragebait and let users who don't want it filter it out or simply recommend based on their voting.