Can Websites Get Depressed?
Can social media sites go through large scale emotional trends? Like, can Twitter or Facebook get depressed or have a manic episode? Can Tumblr as a whole wake up on the wrong side of the bed?
We recently tried to answer that question, at least for BuzzFeed, by using numbers pulled from our reaction bar. BuzzFeed’s team of data scientists were given time periods of major news stories — controversial ones from the last year that caused an uproar of some kind — and they measured positive and negative reaction use for each story.
The question we attempted to answer was whether or not a particularly controversial news story could cause an overall increase of negative reactions across our entire website. Were users in such a rotten mood after reading hard news that it affected the way they reacted to other stories? And does that add up in such a direct way that it’s a quantifiable mood? Think of it like the online version of when you burn your hand on the stove and end up shouting at your dog about it even though it’s not the dog’s fault you accidentally grabbed a hot skillet.
The answer, fascinatingly, is yes. BuzzFeed can become depressed on a macro, sociological level. Which should be a no brainer, but it’s kind of a big deal.
Most social websites follow the Facebook model of like, share, and comment, which at its inception was revolutionarily articulate. Users are typically given two options: to spit the story back out and reshare it, or to simply show their interest or disinterest with some form of like or dislike. Reddit has an upvote/downvote system that essentially shows approval which then moves content; Tumblr has like and reblog; Twitter has favorite or retweet — the list goes on forever.
BuzzFeed is a bit different in that we have an ever-changing reaction bar of various emotions you can assign to posts — custom reactions you can win, a heart for liking, and a broken heart for disliking. If enough people use the same reaction in a quick enough period of time, say “LOL” on a story about a tortoise named Bryan that wears a chef’s hat and lives in a kitchen, it goes into LOL Feed for further promotion.
These sharing buttons and liking options all kind of represent one key thing: For the first time since humans started writing to each other, we’re able to measure and analyze fairly precisely how people emotionally respond to different content.
And those share and like metrics bring up the larger question of whether or not those likes and shares and OMGs and CUTEs mean anything in a greater sense. What it forms is basically a social media economy. You have a picture of a cat with a melon on its head worth 1,000 likes and it’s currently trading on your Facebook against a YouTube video of a Japanese commercial featuring a cat eating a watermelon worth 500 likes and 50 reshares.
For our reaction bar experiment we chose news cycles that caused an uproar in the comment sections. Now comments aren’t the most accurate predictor of how people feel about content, but it’s a good start.
We looked at the reaction bar use a week before a news story broke, the week during our coverage, and then the week after. The percent change every time grew in a way that very closely mirrors BuzzFeed going through a “grumpy mood.”
Time Period: January 9th – January 16th
First frontpage post: Beyonce And Jay-Z Prevented Families From Seeing Their Babies
Tag: Blue Ivy
Blue Ivy Carter, daughter of Jay Z and Beyonce, being born was the first news cycle from this year we used in our survey because it had a lot of elements that make it a perfect controversy on the internet. It’s about celebrities having kids, it involves a wealthy rapper — I could tag this post Lil Wayne, Kanye West, or Tyler The Creator and immediately get an extra 10 thousand hits because people on the internet just by default hate rap (also a lot of people are racists, let’s face it). It was also chosen because the story covered a incident where Beyonce and Jay Z shut down access to an entire post natal care ward because of paparazzi trying to sneak in. So there was a palpable sense of outrage about the rights of the rich versus normal parents.
The percent change in negative reactions from the week before vs. the week during coverage of Blue Ivy’s birth was a sizable 31% increase in negative reactions, with people slamming down on the trashy and wtf buttons especially often.
The next time period we looked at was Whitney Houston’s death in February. Interestingly, Whitney Houston not only increased negative reaction use 38% amongst users, it actually decreased positive reaction by 10%. And like clockwork, negative reaction use dropped back by 29% a week later, with positive reaction totals jumping back to where they were before Houston died by 10%.
This is seemingly common sense, but also really amazing. It’s always been understood that content makes certain people feel certain ways — that’s kind of the point of it. But the idea that a news story can have a measurable, emotional worth of +0% sadness and -10% happiness should creep you out a bit. Whitney Houston death coverage literally brought down the mood of our whole frontpage, like a party guest that pees in your sink.
Time Period: March 22nd – March 29th
First frontpage post: Diverse And Peaceful, A “Million Hoodie March” For Trayvon Martin
Tag: Trayvon Martin
Easily one of the largest news stories of the first half of the year, Trayvon Martin’s death was not only a heartbreaking story about a teenager being killed, it involved strange police reports, very controversial racial undertones, and went on to inspire a grassroots campaign of activism and protest. There were a million ways to cover the story, and BuzzFeed editors followed it pretty closely from a bevy of angles.
Statistically, the reaction buttons followed as they logically should have, with a few curious differences. Negative reactions increased by 26% during our coverage. The interesting thing was that our other reactions, ones that are neither positive or negative, “geeky,” “get real,” and “mamma mia!” the week after we stopped covering Trayon Martin’s death, increased by 154%. Now that says two things, users love the novelty of new reaction buttons, but it also says that as a community, BuzzFeed users were sick of hard news, positive or negative.
Time Period: July 20th – July 27th
First frontpage post: 12 Shot Dead At “Dark Knight Rises” Screening In Colorado
Tag: Aurora Shooting
The next two news cycles, as you can see from the dates, have a significant overlap, which makes them both more complicated to break down, and interesting to look at.
James Holmes walked into a movie theater on July 20th and opened fire on the audience. BuzzFeed entered full hard news mode, and covered the story heavily that Friday, updating it as more details emerged over the next week.
Unlike the Trayvon Martin shooting, the grizzly Aurora movie theater attack was a far more straightforward breaking crime story. What makes it so interesting for this survey is that it’s the only massive breaking news story on the list, and as such, breaking something that emotionally complex down into FAILs, LOLs, and OMGs is difficult.
Our data shows that negative reactions increased marginally, with only a 4% rise. What’s fascinating though is that positive reactions didn’t increase or decrease, but had a 0% change. The week after our coverage of the Aurora Shooting, however, saw an 18% increase in negative reaction bar use, and a 5% increase in positive reactions.
What does all that mean? To answer that we need to look at what happened next.
Time Period: July 23rd – July July 30th
First frontpage post: Mike Huckabee “Incensed” By Criticism Of Chick-Fil-A
Seemingly out of no where, a massive controversy broke out, led largely by Tumblr communities, over the CEO of Chick-Fil-A’s anti-gay marriage stance. His thoughts on gay marriage, and the religiousness of Chick-Fil-a’s owners, mind you, had been public knowledge for years.
We covered the story as Chick-Fil-A had a social media meltdown, and our community rallied around it accordingly, with positive reactions increasing 20% and negative reactions rising by 36%.
Now you could argue that a rabble-rousing attack against the religious beliefs of a fast food company were the result of a bunch of teenagers on Tumblr discovering a known fact and blogging about it, or you could take a step back and argue that the social web actually went through a bout of psychological displacement.
Users couldn’t properly express the anger and sadness they felt over the Aurora shooting story and the social web redirected it at a low-hanging fruit. This is not the first time the social web has done this.
OK DUDE WE GET IT SHUT UP AND TELL ME Y ALL OF THIS MATTERS AND Y I SHUD CARE BCUZ LIKE PLZ I HAVE PIX OF SURFING DOGS OR SOMETHING 2 LOOK AT! GAWD!!!1
This is important because what we’re seeing is the evolution of how people socialize on the internet. Three other stories we used in our survey were Daniel Tosh’s rape joke, our coverage of Coachella 2012, and Obamacare being upheld. Those stories, and the five explained above, all elicit the same kind of “ugh, eff this noise” response. Even the Trayvon Martin shooting had a focal point for anger.
Anger and happiness are emotions that are, and have been, easily expressed online. A great example being the amount of money raised for that bullied bus monitor. But as you can see with Whitney Houston’s death or the Aurora Shooting, we’re now beginning to see social media sites learn how to deal with sadness and more complicated emotions at a macro level.
BuzzFeed’s community, like most online communities, doesn’t deal with it particularly well; it gets moody, sometimes lashes out, or maybe it goes into a slump and everyone begins to feel like the party is over.
So next time you feel like “man, twitter sure is a ghost town today,” or think “Reddit is really depressing, what’s the deal,” know that there’s a pretty good chance that it’s not just you thinking it, but everyone else using the site.