A friend might retweet on Twitter content they were dismissive about, but that retweet would sometimes accidentally legitimize that content by removing the dismissive context (spreading it away in readers' timelines).
Similar with FB showing you stuff friends react to or comment on; their reactions might be Anger and their comments critical, but that context also is still partly lost if you weren't looking for it, and sometimes intentionally whitewashed ("x friends reacted to this" instead of "10 friends hate this, but one kind of liked it"), in the maelstrom of the FB timeline algorithm. (Obviously the worst case on FB up until recently was how easy it was for people to whitewash paid ads by resharing them, losing the context of who paid to promote that content in the first place.)
But the systems optimize for positive interactions (which generates eyeballs/pageview/ad impressions). So even if a dozen friends dismissed something, correctly, that one dumb Shitposter will get the word accross and the sytem will benefit the thumbs up with a zero effort propagation (one click) versus the critic (time to sit down, write an argument, work to have that thumbsuped... or o HN: create an account years in the past, acumulates karma just so you can downvote things that are actually wrong).
the incentives are the problem (advertising) not presence of wrong content amidst good content.