Long thread - buckle up. TL:DR; yes, you should join Mastodon. But you should stay on Twitter as well. What we need are more and different online communities, not just an exodus from a troubled platform.
I opened an account on Mastodon in 2017 as part of my research on a fascinating experiment: an open source, decentralized Twitter alternative. There was a moment in 2017 when it looked like there might be a serious exodus from Twitter to something better. That moment was brief.
By the summer of 2017, my US and European geek friends had largely forgotten Mastodon. I discovered that most of the growth of the network had been from Japanese lolicon fans and wrote a (controversial) essay about the platform's growth: ethanzuckerman.com/2017/08/18/mas…
I learned two things from Mastodon and lolicon: 1) federation is actually a terrific way of handling content that some communities find acceptable and others don't, and 2) you learn a lot from studying non-mainstream groups and how they move between online spaces
I've gotten more new followers on Mastodon in the last week than I have in the previous five years. Evidently people are a little nervous about this Musk guy and what might happen to this platform. They're right to worry, especially as Musk now seems to be gutting the company.
Musk has already used his massive reach to promote partisan disinformation: npr.org/2022/10/31/113…
And a wave of trolls is testing whether Twitter will actually enforce prohibitions on hate speech. washingtonpost.com/technology/202…
But here's the thing: while I think everyone should try open source, decentralized networks as an alternative to Twitter, I also think this is the wrong time to leave Twitter. I'm going to lean on economist Albert Hirschman to explain why.
In 1970, Hirschman wrote a brilliant book called "Exit, Voice and Loyalty". It examined an apparently simple question of microeconomics: What should you do if a product you use gets markedly worse? (Filled with fascist conspiracies, for instance...)
Classical economics says you switch to a competitor who's got a better product at the same price. But not everyone acts like a rational consumer. Sometimes people tolerate poor performance because of loyalty to a brand. And sometimes they raise their voices.
If you think you might be able to influence the company and feel enough loyalty, you might choose to use your voice and complain about the change in quality either as an alternative to exiting or as a step before you exit.
Near the end of the book, Hirschman proposes a special case: politics. You can't exit a state, Hirschman argues, so voice is your only tool. Even if you put your kids in private school, your taxes still pay for public school. So no real exit is impossible.
(Musk's former PayPal founder Peter Thiel has been associated with all sorts of efforts, from space colonization to cryptocurrency to seasteading that seek exit from existing governments. These projects tend to ignore how interdependent humans actually are.)
While Musk might exit for Mars any day now, you can't easily exit Twitter, much in the same way that you can't easily exit your home nation. Even if you leave - even if a whole lot of us leave - Twitter will have a special and influential place in the world.
Because Twitter made writing public by default - unlike Facebook - it became a cheap and dirty way for journalists to put public opinion into their stories. No, Twitter's not an accurate sampling of public opinion. But it's allowed Twitter users to be highly influential.
Twitter can be a powerful amplifier for views we don't often hear, especially if those views are interesting, controversial, snarky, or funny. From #OscarsSoWhite to BLM, Twitter has often platformed marginalized voices and amplified them via conventional media.
It took a long time for Twitter to become journalism's watering hole. Maybe Mastodon will eventually fill that niche. Until then, leaving Twitter means ceding the platform to those most attracted to Musk's policies and politics, and giving them amplification power.
If Hirschman doesn't convince you to stay, here's another argument: Twitter has a great architecture to break echo chambers. Since you can follow whoever you want, without a need for a followback, it's possible to curate a highly diverse network of voices to listen to.
It's not a guarantee - you can certainly let Twitter become an echo chamber. But unlike on Facebook, it's possible to follow people you're really distant from and learn from them. Until Mastodon moves beyond a pretty elite set of early adopters, that's hard to replicate.
So yes, please do create a Mastodon account. Use instances.social to choose an instance whose rules work well for how you want to communicate. Read those rules carefully and make sure you trust whoever controls your instance.
But then - crosspost. Use a service like crossposter.masto.donte.com.br or moa.party, which echo your tweets as toots, and vice versa. Follow the conversations on both platforms. For me, they're quite different.
But that's not all you should do. Consider setting up your own social network for a small group of friends. My team at UMass is running one for our lab and for a group of close collaborators. It's a team space that's less insistent than slack.
Darius Kazemi (@tinysubversions) has a version of Mastodon that he's optimized for small circles of friends, called Hometown. github.com/hometown-fork/… He's got excellent documentation and start up information.
(Our lab's product Smalltown has learned a lot from him - our product is optimized for small civic communities, rather than friend groups. We are using our own Mastodon fork for our internal conversations.)
Why Twitter and Mastodon and even running your own social network? Because there needs to be more than one kind of space. Think about the physical world. A church, a library and a bar are all rooms. But they've got different affordances, and different norms.
It's hard to get a beer in most churches, and it's often too loud to read in a bar. But each time you enter that room, you have a basic understanding of what the rules of the space are, and how you might conduct yourself. Twitter is often just one big room.
The big room has advantages - you hear interesting voices you otherwise wouldn't. But we need smaller, special purpose rooms as well. A political conversation between strangers might demand careful moderation, a chat between friends does not.
In the long run, we need a variety of online social spaces with a wide variety of affordances, purposes and norms. Reddit is a step in the right direction, with different rules in different subreddits.
So are special-purpose sites like An Archive of Ones Own, which is designed and run by fanfiction authors who got sick of bumping into rules and moderation decisions that didn't make sense for their community. (archiveofourown.org)
We need tools that will let us manage all these different rooms from a single app on our phones. That's coming - our Gobo project will first let you read and post to Twitter and different Mastodon-like networks from the same client.
The next release will add the ability to pick and choose between algorithms to sort the content you're subscribed to. We hope to show people that managing your relationships with multiple platforms can, in the long run, be easier and better than having one you don't control.
For now, see you on Mastodon (@ethanz@octodon.social) And Twitter. And Smalltown and Reddit and Facebook for that matter. We need more conversations, not fewer.
• • •
Missing some Tweet in this thread? You can try to
force a refresh
Paul Meosky, a doctoral student at Yale, presents fascinating research on encouraging civil conversations on Nextdoor - a deep interest of mine and @chandrn_ . The idea was to move conversations from Nextdoor into groups with an architecture designed to create civility.
There were carefully stated guidelines, designed around the principles of procedural justice. Within this architected space, civil discourse and "virtuous" speech outweighed problematic speech. Reports of misuse of the platform decreased significantly.
The takeaways are straightforward: design can promote civility and reduce incivility. The problem: "civility" is a slippery concept, and understanding what we're aiming for is complex. How can spaces design not just to minimize uncivil speech but enable civic speech?
Amazing paper from Beth Goldberg, Rick Sear and Yonatan Lupu (@yonatanlupu) on detecting and analyzing "hate clusters". They've found ~1900 communities in their analysis of VK, Telegram, 4chan, insta, gab and FB (since 2019) and Bitchute, Rumble, Twitter and YouTUbe since 2021.
Based on a ton of handcoding, they've trained ML classifiers to analyze 7 different types of hateful speech and 4 problematic movements, including QAnon and Boogaloo. They're analyzing in terms of political and platform shocks.
After George Floyd's death, they see an increase in hate speech, not only around race (3x) but around gender identity (2x) and other hate categories. What's particularly interesting - after the spike of Floyd's death, the base level of race hate speech is higher for many months
Okay, like virtually everyone else who studies social media, I'm fascinated by @elonmusk's takeover of Twitter. Let me say upfront: I have no idea what Twitter will be like under Elon's leadership and neither do you. But that's the point.
We already know little about how Twitter handles content moderation or how their algorithms work, and taking the company private makes it likely we will know less. What we do know is this: two billionaires will now control four of the major digital public sphere platforms.
Zuckerberg, by virtue of his unusual "founders shares", has functional veto power over much of the decisionmaking within Facebook, Instagram and WhatsApp. (morningstar.com/articles/10612…) A privately held Twitter will likely give Musk similar personal power over Twitter.
Conor Friedersdorf is right to point out that fighting mis/disinfo has become an industry, and sometimes a mindset... and that the assumptions behind that mindset aren't all correct. I was at the same conference and also worried about emerging groupthink.
If I were writing this piece - theatlantic.com/ideas/archive/… - my focus might have been on the limits of regulation to solve these problems. In other words, for all the enthusiasm of Senator Klobuchar and others, much of our current information disorder can't be regulated away.
Instead, it's become mainstream political speech. The problem, alas, is not algorithmic amplification so much as it is ensuring there are consequences for political speech that is false and dangerous. I wish that were as easy to solve as writing ML code to detect it.
I got an interesting email from a young Nigerian today alerting me to the problem of social shaming and microlending. He wanted me to promote his petition - change.org/p/central-bank… - and so I did a little research. (1/n)
App-based microlending is popular in Nigeria, both because banks generally require collateral for loans and because the bureaucracy involved with lending can be overwhelming. So there's a flock of new lending sites springing up, like FairMoney. (2/n)
FairMoney reported lending out $93m USD in 2020 in amounts from $3-$1000. Interest rates ranged from 30-260% APR, against 15-20% APR for bank rates. In other words, these are payday loans with predatory terms, but they're what people can access. techcrunch.com/2021/02/18/wit… (3/n)
Like Jimmy, I remember the web before section 230. At Tripod.com, we were providing homepage hosting services to thousands of users. Like Geocities, our main competitor, our product was free to use, supported by advertising.
We needed millions of users to make the economics of the product work, and we needed our support costs to be fairly low. Reviewing every piece of content before it was posted would not have been possible - we simply would not have offered our homepage building tool.
People figured out clever ways to use the server space we provided to share pirated software. As is typical in these cases, the users were often one step ahead of us - they'd learn how to mislabel a .rar file as a .jpg and evade our automated tools.