In my area Secret had a similar story - started out being popular with people in their 20+'s, putting out some insightful/funny content (perhaps with some light local context). Content slowly started spiraling downward toward very specific sort of circle-jerked inside jokes (e.g. jokes about some high-schools, fraternities and even specific kids at larger schools).
Anyway it is a shame because there were people out there put out genuinely good content.
The experience is similar with the app 'Fling' (Send snapchat-like picture to 50 random people) 70% of the content is guys lying in bed with 'lol hi any gurls?'.
Perhaps a way to remedy this would be to cluster on 'like-mindedness'. Then have content show more according to whatever cluster you are in. Constant feedback on what you like and what you put out.
Some good ML + NLP based filtering could give the impression at least that many like-minded people are using the app and inspire people to put out content. If you don't allow users to control the content they follow (Facebook / twitter / vine / youtube / snapchat - 'supervised') you have to direct the right kind of content to users in an 'unsupervised' way.