Google has done a terrible job, but perplexity has done a pretty good job. I think there's quite a bit to be gained by summing/extracting desired information from threads. I'm thinking opinions on choosing tech stacks (eg, "should i use sql or nosql", "fastAPI vs supabase"...), or even fixing bugs (on that note, if you're going after tech, github might be another to look at).
I defitley agree there's a ton of noise in discussions, I think LLM's are the current best way to cut through that noise.