How often do you want to be terrified? Anxious? Sad?
How often do you want to collapse into full-body panic, or curl up into a ball and cry, or get so angry you can’t keep any other thoughts inside your head besides the thing that must be destroyed—even though you don’t have the power to destroy it?
I think that the answer, for most of us, is less often.
Caveats are in order. Most of us don’t go out and seek these emotions on purpose. If we do, it’s normally in controlled environments, such as movies, that are designed to be time-limited and to end with a scheduled cathartic release. The rest of the time, we’re simply hit by life, which often chooses our conditions for us. Someone who comes to you promising a foolproof cure for all of your negative emotions is selling either snake oil or fatal poison.
These emotions also have functions. It’s no accident that they’ve lasted millions of years to affect brains today. Emotions can function faster, if not always better, than rational faculties. Anxiety, for instance, might give me a heads-up to some area of concern I’d otherwise have missed. If I’m stuck in a depressive stupor, fear or anger might spur me to necessary action.
But fear, anxiety and rage don’t equal, and can’t replace, plans of action. My rational brain should eventually take over and make such a plan to deal, if possible, with the actual causes of these emotions. That’s done, I’ve no need to continue scaring myself worse and worse, or whip myself up into a frothing rage. If I did so anyway, those negative emotions could severely hamper my attempts to actually execute whatever plan of action I’d chosen.
A bit more advanced is the idea that fear, anxiety and rage beyond one’s capacity to cope can thus very well loop around and cause more of themselves. But it’s pretty easy to see if the mechanism is explained. Unceasing emotional turmoil—whether expressed or, perhaps worse, repressed—becomes a problem for others to deal with; if the people around that person aren’t total saints or oblivious rubber people, they’ll be affected by this turmoil. They’ll struggle to treat the sufferer as well as they’d like to; the sufferer will take note of this, and react in a variety of ways, none of them good. In short, the negative emotions lose their original referents, and become a self-fulfilling prophecy.
I’ve seen this happen many times. It can wreck your entire life.
But even if you avoid it, and no one around you loses faith, that’s no consolation. Feeling constantly terrified, or sad, or hopeless, will still sap your soul.
Social media use is strongly correlated with the complex of symptoms known as depression. A study by Liu yi Lin and Jaime E. Sidani, et al., measured those symptoms using the PROMIS Depression Scale Short Form and self-reported use frequency over 11 social media sites, including Facebook, Twitter, Tumblr, Reddit, and Snapchat. That study found that the top 25% of frequent social media users in their study population “had significantly increased odds of depression […] after controlling for all covariates.” 
Depression sucks. Obviously. If it were as simple as “social media use causes depression,” this argument would be over.
It’s not even close to that simple.
First, correlation doesn’t imply causation. A hidden variable unknown to the study’s authors could be causing both increased social media use and depression. Second, even if there is causation, the direction of the causation isn’t clear. Does something about looking social media cause more depression, or does depression cause people to distract themselves by spending more time on social media? The Lin/Sidani study explicitly refuses to conclude one way or another. 
The study does not argue that if a depressed person quits social media, they’ll stop being depressed. It also doesn’t argue that social media caused the depression, which may in any case have been pre-existing.* Lin and Sidani, et al. do throw out four potential explanations of the correlation, only the last of which clearly differentiates the Internet from, for example, television. 
That last is online harassment, which is a very serious problem—and much worse if you’re a millennial. But cases bad enough to provoke serious, long-term negative emotional reactions just don’t occur often enough to explain the correlation.  Besides, people can be, and are, harassed on message boards, via email, in comment sections, and so forth. Harassment isn’t a problem unique to social media, just one that appears most often there due to the immense popularity social media enjoys.
It’s unfortunate that an up-to-date study from just last year, extensively citing the relevant literature, can’t even guess at an explanation of its own findings about social media that is actually unique to social media.
But such explanations do exist. I’ll offer one below.
Here follows a story about Twitter, the social network I’ve used the most over the last year.
Twitter is struggling financially, and has never turned a profit. Its shareholders would like it to continue to exist, though, so Twitter needs to aggressively maximize revenue. Twitter’s major historical source of income has been advertising on the platform in text form.
Text is useful for advertising only if customers actually read it. Twitter solves this problem with “sponsored tweets,” which—ideally—blend almost seamlessly with real tweets.  The idea is that you’ll see their ads just in the act of reading your Twitter timeline; they’re not off to the side, but embedded within the main feature of the platform.
In short, to ensure its long-term survival, Twitter needs make sure that its customers come to their feed frequently enough, and stay online long enough, to see, read, and perhaps engage with these sponsored tweets.
At almost all costs.
Twitter has every reason to utterly ruin their service, if doing so gets clicks on those tweets.
What would stop them? Twitter’s top decision-makers don’t rely on Twitter very much in their daily lives. Perhaps they would like to believe that Twitter is a fun thing to use heavily, but they can only know this second-hand.
Jack Dorsey, the founder and CEO of Twitter, produces only a few tweets a day. Many of them are just PR: innocuous if barely meaningful statements like “New markets: we aspire to be a global service.” When he bothers to interact with other human beings at all, those interactions are about Twitter.
Dorsey does not, in short, use his public Twitter account to have real human conversations. He has those, I’m sure—they’re just located somewhere that is not the social network he personally runs.
That should, perhaps, be a warning sign.
I’ve noticed that, on Twitter, the upper-tier of the very most intensely emotional threads and RTs and Storifies and external links get copied around dozens of times, with minor variations in commentary, inevitably showing up in my timeline over and over again.
That’s a problem, because—when, as often, those stories provoke fear, anxiety, and/or rage—it’s really only useful to see them once, perhaps twice as a reminder, before formulating a response and, if necessary, a plan of action.
Seeing them repeatedly echoed is redundant; seeing them repeatedly echoed in slightly different words is worse, because then you need to parse the story all over again. Either way, subsequent impressions won’t do you any good, but you’ll still probably feel at least a bit of fear, anxiety, and/or rage.
There’s no technical reason stopping Twitter from collating external links posted by more than one account you’re following. (Facebook already does something similar.) If Twitter did that, then made it handy to mute the story right from that tweet, then using Twitter would become considerably less distressing.
Sadly, I doubt Twitter will implement that feature anytime soon.  Fear, anxiety and rage are too effective. Stoking them keeps you staring at the screen; it keeps you interacting with the site so you can blow the steam off.
It keeps you seeing advertising.
This kind of assault on the psyche is pretty much bound to tick up the rate of illnesses like depression, by at least a few percent. It could very easily explain the results of the Lin/Sidani study.
But it’s worse than that.
Your negative emotions—the ones you hate, and would like less of—are, in fact, part of the life support sustaining Twitter’s very existence.
That’s Twitter. But I think that story rings familiar for every other big social media site. In the current environment, that’s just what social media is. It’s advertising driven; one way or another, it needs to keep people there, and looking. The good of such sites’ users will inevitably become, at best, a secondary priority.
One can, of course, know all of this and continue to use sites like Twitter. It’s defensible, in fact. You can legitimately meet interesting new people on social media, which has long been its selling point. If you have friends who won’t use anything else, or if your audience is located there, then leaving any given platform entirely might be a significant sacrifice.
On the other hand, you might be best off evacuating social media completely, deleting your accounts and ripping off the rear view mirror. It’s really up to you.
For the rest of you, I have a simpler request: just don’t build a nest there. Don’t check Twitter first thing in the morning and last thing at night. Reconsider using multi-account platforms like TweetDeck, which only amplify the risk of overload. Above all, don’t make checking your Twitter timeline or Facebook feed the default action when taking a break. Social media isn’t built to be the place you return to after an exhausting journey, even if the people there are ones you’d like to see.
Instead, use that time—that should be used for emotional decompression and mental relaxation—to enjoy or participate in anything not explicitly financially incentivized to hurt you.
You’ll be better off for it.
 Lin, Liu yi; Sidani, Jaime E.; et al., “Association Between Social Media Use and Depression Among U.S. Young Adults.” Depression and Anxiety 00:1-9 (2016.) 1-3. This article is available freely online.
 Lin, Sidani, et al., 6. “Because our data were cross-sectional, the directionality of this association is not clear.”
 The other proffered explanations are: envy at seeing idealized/carefully curated representations of others and comparing those representations to their own known inadequacies; disgust at perceiving one’s own time on social media to be wasted; and unexplained “Internet addiction.”
 Duggan, Maeve. Online Harassment. Pew Research Center. Only about 6% of all Internet users reported having experienced harassment that was “extremely” or “very” upsetting. (I’m in that 6%. Said harassment was, nevertheless, not sustained; it affected my social media use, but its effect on my mood disappeared not long after the harassment itself did.)
While I’m being anecdotal, I should also say that when I have been witness to sustained harassment, the usual effect is for the person to leave the platform entirely. Such responses appear in significant numbers in the Pew Research survey, though their correlation to sustained versus acute harassment isn’t explored.
If those who do experience sustained harassment generally leave the platform, it’s notable that they would then would be counted afterward as non-users of social media in the Lin and Sidani study. So, even if their depression lingered after sustained harassment, they wouldn’t be part of the statistically significant increase in depression among heavy social media users in that study.
 In practice, the effect is often jarring. An average person’s twitter feed might well have lots of corporate spam just by virtue of accounts that person follows voluntarily—sports accounts, for example, which often cross-promote with a variety of businesses. A millennial’s pseudonymous twitter, however, or an artist’s art-posting account, aren’t likely to follow anything remotely resembling Twitter’s usual advertisers.
 If Twitter does implement that feature, I’d predict that it was only after finding some new way to aggravate its users. For instance, Facebook’s constant push of its tagging feature, leading to lots of unexpected arguments with random upset strangers, more than undoes whatever good its link-collation feature might have done.