The thing about algorithms is they’re not very good at reading a room.
When algorithms on Twitter, for example, see increased activity around a certain phrase or hashtag, they will typically assume that piece of content is worthy of extra attention. Before you know it, the hashtag appears in the trending topics sidebar, which begets even more attention, which creates a snowball effect that can leave that particular hashtag in the trending sidebar for hours and hours and hours.
The next thing you know, media outlets are writing articles and airing news segments about how that hashtag was an actual thing, which amplifies the ostensible sentiment behind the hashtag, often without context or background about why that hashtag was trending in the first place. This is problematic because hashtags trend on Twitter for all kinds of reasons, and often those reasons are antithetical to the apparent aim of the hashtag itself.
Such was the case today when the hashtag #NeverWarren was trending on Twitter for much of the morning and afternoon. It appeared after the Democratic debate in Des Moines last night, during which many supporters of Bernie Sanders became upset with Elizabeth Warren after a few on-stage disagreements and a seemingly very uncomfortable exchange after the event. Even 12 hours later, the hashtag was still one of the top trending topics on Twitter.
Some Sanders supporters are genuinely upset with Warren, but as many commentators pointed out today, that’s not what kept the hashtag going for so long. Rather, it was fueled in large part by people denouncing the sentiment behind it. That’s right. #NeverWarren was the product of a lot of people saying #NotNeverWarren.
FASCINATING data here about #NeverWarren, the No. 1 trend in the United States right now.
The algorithm is generating an opposite reality. pic.twitter.com/lAOyiS5S7G
— Ben Collins (@oneunderscore__) January 15, 2020
Still with me? All of this gets back algorithms and their inability to read a room. Understanding sentiment on the internet is hard enough for humans without automated systems amplifying content for the wrong reasons. Factor in bots, trolls, irony, and any number of bad-faith variables, and it’s easy to conclude that the best response to a trending hashtag is often to just ignore it. Nothing will make it disappear faster.