Twitter's problems: a roundup
Some unsolicited thoughts for the new ownership
So, Elon Musk bought Twitter. Personally, I’m pretty sanguine about this development. It’s no secret that I think that Twitter is a uniquely dystopian feature of the modern media sphere — a bad equilibrium that traps the nation’s journalists, politicians, and intellectuals in close quarters with all the nastiest and most strident anonymous bottom-feeders. There’s a nonzero chance Musk will be able to improve this situation; if not, it’s hard to see how he could make it particularly worse. If he destroys the platform, we’ll find something else — probably a number of different somethings, which I think would be good for the media ecosystem. Our entire society was not meant to be locked in a single small room together; we need more room to spread out and be ourselves. (And personally, I’ve already ported as many as possible of my Twitter followers over to my Substack email list, so I’m not particularly worried that my “brand” will be devalued should Twitter go down the tubes.)
But anyway, I am optimistic that Elon might find some way to improve the “hellsite”, as many of its users call it. At least some of the people advising him are good people who know their stuff. So I thought I’d do a roundup of the various posts I’ve written about Twitter over the years, as well as one that I didn’t write.
1. The quote-tweet button
I’ve known Chris Wetherell for a long time, and he’s one of the kindest human beings I’ve ever met. Among other things, he’s also the inventor of the retweet button. A 2019 Buzzfeed article chronicles how Chris saw the feature he invented — in particular, the quote-tweet feature — immediately used to form harassment mobs that terrorized some of his own friends:
Jason Goldman, the head of product when Wetherell built the retweet, said it’s a key source of Twitter’s problems today. “The biggest problem is the quote retweet,” Goldman told BuzzFeed News. “Quote retweet allows for the dunk. It’s the dunk mechanism.”…
[T]he [quote-tweet] button…changed Twitter in a way Wetherell and his colleagues didn’t anticipate. Copying and pasting made people look at what they shared, and think about it, at least for a moment. When the retweet button debuted, that friction diminished. Impulse superseded the at-least-minimal degree of thoughtfulness once baked into sharing. Before the retweet, Twitter was largely a convivial place. After, all hell broke loose — and spread…
As [the] Gamergate [controversy] unfolded [in 2014], Wetherell noticed its participants were using the retweet to “brigade,” or coordinate their attacks against their targets, disseminating misinformation and outrage at a pace that made it difficult to fight back. The retweet button propelled Gamergate, according to an analysis by the technologist and blogger Andy Baio. In his study of 316,669 Gamergate tweets sent over 72 hours, 217,384 were retweets, or about 69%…
Gamergate was a "creeping horror story for me," Wetherell said. "It dawned on me that this was not some small subset of people acting aberrantly. This might be how people behave. And that scared me to death.”
Twitter, from that moment, became an “anger video game.” Retweets were the points.
Gamergate is long gone, but the basic format that it pioneered — big accounts using quote-tweets to call in mobs of followers to attack someone they don’t like — became the standard on Twitter. From liberals and leftists to conservatives and rightists to nationalists and partisans of every conceivable cause, Twitter has become a battlespace of warring gangs that use quote-tweet mobs as their artillery.
So far, the cautious people who run Twitter have been unwilling to mess with this golden goose — quote-tweet mobs are, presumably, a big reason why people stay so engaged with Twitter. But Elon, being more of a bold experimenter, may be willing and able to fix the platform’s core feature.
2. The Shouting Class
In this post, originally published at my old blog in 2017, I explained what I think is the other core reason why Twitter is such a bad influence on our society. Essentially, the platform suffers from a gigantic negative selection bias — unlike Instagram or Discord or Snapchat or even Facebook, Twitter is biased toward the people with with the greatest inclination to shout and complain and fight. This includes trolls and sadists, divisive ideological agitators, foreign government operatives, and people with emotional problems who use Twitter battles as a substitute for therapy.
Before Twitter, these people would mostly be bothering their families or colleagues, writing angry comments on news stories or letters to the editor, or engaged in highly localized mischief. Society naturally contained them. But Twitter flipped the script; thanks to Twitter’s quote-tweet feature, these people can win fame and even fortune by taking their toxicity public. And thanks to Twitter’s radical openness, these shouters appear more numerous than they really are — when the same 100,000 people go around doing a ton of angry replies all day, journalists and politicians and intellectuals can easily come to believe that the shouters are representative of broader society.
In a follow-up post, I cited some academic evidence that Twitter attracts and elevates hostile individuals who are drawn by the promise of social status, and how this biases Twitter discussions toward outrage:
A hellscape, after all, must be populated by pitchfork-wielding devils.
Fixing Twitter will therefore require figuring out how to bias the internet’s public square away from the toxic minority and back toward the reasonable majority.
3. Cancel culture
In addition to shouting and mobs, Twitter regularly features attempts to pressure organizations — companies, universities, nonprofits, government agencies, etc. — into obeying the will of a small activist minority. In a 2020 Bloomberg article, I described this as a form of “gekokujo” — borrowing an old Japanese word that was used to describe how ultranationalist junior officers used public support to overawe their superiors back in the turbulent 1930s.
Imagine you’re an administrator at a university. You make a mistake and hire a faculty member who does nothing but start fights in the department. So you do what comes naturally — you fire that person. In the old days, that would be that. But now suddenly you find yourself deluged with a hundred angry comments from Twitter users who aren’t even affiliated with your university, demanding you reinstate the fired faculty member. Terrified by the power of negative popular attention, you reinstate the disruptive faculty member.
In fact, this is probably the wrong response. The 200 random people who screamed at you on Twitter were probably exactly that: random people. Bored at work, or dedicated to some political cause, or simply unhappy in their own lives, they decided to joining the angry mini-mob. There’s vanishingly little chance that they’ll actually be able to organize a popular boycott or impose any other sort of direct consequence on you or your organization. And 200 people is really a tiny number of people. But our brains are conditioned to think that if 200 people are yelling in our face, we’re in grave personal danger. So people end up backing down, even when they really don’t have to. I wrote a post about this last year:
Twitter’s power of “cancellation” is boosted by another feature of the internet: permanence. Anyone who searches for your name will always be able to find those 200 people who got mad at you on Twitter. This means that “cancel culture” is different than traditional social ostracism, which you could escape by moving to a new town or by people forgetting a controversy over time:
Anyway, I’m not going to claim that Twitter-driven cancel culture is a purely bad thing; I think it has done great good in some areas of our society, providing a natural check on the power of hierarchical closed-off organizations. But a society does need stable organizations and hierarchies in order to function, and there are clear signs that cancel culture has paralyzed a lot of orgs — especially progressive ones — into ineffectuality.
Fixing this aspect of Twitter probably just requires fixing the quote-tweet feature and disempowering the Shouting Class. But it adds urgency to the need to fix those other two problems.
4. Status anxiety
There’s one more problem I see with Twitter, though this one is probably less socially important than the others. Twitter has a way of evoking weird status anxiety. Smaller and unverified accounts often feel resentful of larger accounts and verified “bluechecks”. This resentment is actually amplified by the platform’s universal accessibility — you can talk directly to a 4 million follower bluecheck, or even dunk on them, but then at the end of the day you don’t have anywhere near their reach and platform. As I wrote in a post last year, I think this sense of constantly revoked pseudo-egalitarianism exacerbates all of Twitter’s other issues:
This is probably the last thing I’ll write about Twitter for a good long while (which will come as a relief to readers who feel like I harp on the subject too much). In all of these cases, I feel like I’m good at diagnosing the problems but don’t have obvious solutions in mind. Twitter is a platform whose use cases and culture evolved rather than being planned out in advance, and fixing it will require doing a lot of the hard planning that the initial creators neglected. It will probably require experimentation, observation, setbacks, missteps, and persistent effort — all of which will draw the ire of the Shouting Class who have gotten used to their privileged position in the Twitter-driven world.
But it needs to be attempted, because most of us are getting tired of the bad equilibrium that Twitter has trapped our society in. And if it can’t be done, then hopefully people will migrate to other, less inherently toxic discussion forums.
I think Ben Thompson described an interesting plan for addressing Twitter’s content moderation and financial model on Apr 18 in his article, “Back to the Future of Twitter”.  His basic idea is to separate Twitter into a backend service that hosts content and a front end client that provides moderation and algorithmic curation. Twitter would then open the backend up so that anyone can develop their own client by paying for access.
The backend would do the minimal amount of moderation necessary to conform with the law in each jurisdiction and then provide optional services to the clients for more sophisticated moderation and algorithmic filtering and ranking. Twitter would continue to host their existing front end and possibly create more over time.
The idea is that the plethora of clients would allow for experimentation and meeting the diverse user preferences. You could have heavily moderated clients as well as wild west anything goes. While all clients would reference the same social graph and tweets/replies, each client could use whatever methods they want for filtering and ranking.
Some possible clients I’ve considered.
1. A client geared towards journalists and other prominent people that provides them with a layer that filters out harassment. This could include manual curation of their DMs, replies and retweets. The service would also have a team of lawyers for reporting credible threats of violence to law enforcement and sending cease-and-desist letters for proper libel. (Note this works well with Elon’s plan to require human identity verification, even for anonymous accounts.) This service would be expensive and geared towards people that value this layer of protection for professional reasons.
2. A client geared towards techies like myself that want to experiment with their own algorithms. While we’re a niche group, many of us would gladly pay quite a bit of money to support such a nerdy hobby. There’d be sharing and critique of each other's work as we toil to optimize our own Twitter experience. Over time some of the ideas might filter into other mainstream clients.
3. Partisan clients. Fox news could provide a client as could CNN. Even the DNC, RNC, and other parties could provide their preferred view of the social graph. They’d be openly filtering and ranking the social feed in a way that corresponds to the user's political preference.
There are three possible outcomes
1) Elon makes Twitter better -> a win for society
2) Elon destroys Twitter -> a win for society
3) Elon somehow makes Twitter worse (probably impossible), thus making Twitter users suffer more -> a win for society