Back when I browsed TV Tropes regularly, my algorithm for using it to find media that I liked centered around skimming a lot of media pages that looked vaguely interesting and using them to get a better idea of themes and target audience, while throwing out anything that was full of creepy fanservice tropes or obviously written by a single very enthusiastic viewer. When I tried mining it for actual recommendations, they were usually bad.
LW doesn't have anything that lends itself to that sort of exploratory search, but recommendations from the media threads have been somewhat reliable for me, probably thanks to a closer demographic match. Better coverage in certain topic areas, too: we seem to have a greater proportion of literary SF readers posting here, for example.
Several weeks ago I wrote a heavily upvoted post called Don't Be Afraid of Asking Personally Important Questions on LessWrong. I thought it would only be due diligence if I tried to track users on LessWrong who have received advice on this site and it's backfired. In other words, to avoid bias in the record, we might notice what LessWrong as a community is bad at giving advice about. So, I'm seeking feedback. If you have anecdotes or data of how a plan or advice directly from LessWrong backfired, failed, or didn't lead to satisfaction, please share below.