Begun, the Twitter clone wars have

Threads has garnered a big user base and Mastodon chugs along. Now Bluesky is officially open to all. Finally some real competition for Twitter refugees.

Regular readers of this space know I’ve spent considerable mental energy and spilled considerable digital ink wondering what to do about Twitter since the sale to Elon Musk. I’ve been in lament mode for a while, knowing that even as people were leaving it was important to acknowledge that something significant is lost when a community deteriorates due to Elon’s particular skill at stepping on rakes. He’s deteriorated network trust with his crazy verification gambit, which by the way isn’t working since it’s earning the company a pittance all while giving the signal boost that comes with a paid checkmark to … bona fide terrorists. If only everyone hadn’t seen this coming.

Since the sale, we’ve seen the rise of several different options acting as a refuge for people who left the network. Mastodon, Threads (I wrote about it here), Post News, Spoutible, Hive, and several other notables.

For the most part, these Twitter clones have been pretty niche. They’ve got their advocates but still are pretty isolated other than Threads, which is Meta’s product that is getting about 130 million monthly active users. Threads’ daily active users (DAU) count that advertisers look at is a bit harder to cut through because it’s based on a lot of conflicting reports, some of which require time travel or deep knowledge of the laws of physics. But I’d estimate it at being around 30 million DAU as a synthesis of what I’ve seen reported. For context, Twitter had something close to 250 million DAU before the sale (though we know that has fallen precipitously since). It’s likely Twitter still is getting significantly more DAU than Threads, but you have to remember Threads is only about 8 months old and has the network effects possibilities of Facebook and Instagram, meaning it isn’t going away. And as I argued in the piece linked above, it doesn’t even need to be the top Twitter alt to win. All it has to do is hurt Twitter significantly, which it is doing.

But the reason I’m writing today is because Bluesky, yet another Twitter clone, has finally opened to all after years of development. When I quit Twitter last summer, Bluesky was where I placed my bet, but at the time it was invite-only and was slowly populated by giving users codes to share. Being finally open is a significant milestone, and the response has been significant. On the first day it opened it got almost 1 million new users, increasing its user base by about 30%.

So what is Bluesky? Interestingly enough, it’s a project that started out of Twitter before the sale. Twitter co-founder Jack has long lamented the decision to make Twitter a publicly traded company, which left the platform beholden to shareholder pressures and the whims of advertisers. So while he was CEO, Twitter funded the startup that made Bluesky possible by enabling it to create a public benefit corporation that could run a Twitter-like network as a non-profit and create a vision for a decentralized large-scale network. I’m still somewhat amazed Twitter’s board allowed this to happen given what has come after.

While in development, Bluesky did two things on parallel tracks.

The first was to build a decentralized, open-source platform software called the AT Protocol. Like any open-source software (think WordPress and its ilk), anyone could take the code, install it, and tweak it to their liking. “Protocol” is a bit of a wonky term, but you already use one when you use something like email. That is, the reason you can email someone from Gmail to Yahoo (i.e. it’s not internal to one service) is because someone created software that lets different services do a handshake and communicate in the same language. AT Protocol is the same vision but for social networks. Twitter is a singular thing, run and owned by one company; anything you can do is internal to its servers and whims. Bluesky’s AT Protocol is open-source software, meaning anyone can create networks with different rules but it gives users the ability to handshake and communicate.

In layman’s terms, you can control your own social network experience just as you control your email experience by who you send to, receive from, share email addresses with, and (crucially) block. More on that in a bit. But think of it as if people on Facebook could see posts on Twitter and be seen on Twitter, and multiply it by all the social networks out there. In that kind of world, the platform you’re on doesn’t matter and it severely limits the power of a particular service because you own your content and your network.

The second thing Bluesky did was create a network it owns and install its software (they call it “the app” that uses the software to differentiate it from the software protocol itself, because confusingly enough both are named Bluesky). Again, this is just one version of what can run on AT Protocol, but it addresses one of the crucial problems early Twitter clones such as Mastodon experienced by giving people a place to join at startup. But you aren’t stuck there. You can spin up a version of the software on your website with rules more to your liking and port your contacts with you, meaning you own your posts, data, and follows/followers. One of the real problems with Twitter was starting over, having to rebuild a whole network again, and losing all your posts. It’s certainly something I disliked about leaving my 16,000 followers on Twitter behind. AT Protocol promises that this is a thing of the past. Unlike death, you can take it with you when you switch servers.

But there is one thing I like about the Bluesky app. It’s fun. Threads can be fun in a normie kind of way, but Bluesky is just a weird kind of fun at times in all the ways the early Twitter was. I was talking to a colleague and I said Bluesky is like Santa Rosa to Threads’ San Francisco, an extremely niche reference. San Francisco has a reputation for being weird but is a pretty button-down city full of corporate professionals. Santa Rosa is how a lot of the U.S. pictures San Francisco in socially constructed form; it’s full of free spirits and artists and doesn’t care what you think.

And I see that split all the time when I log on to the two competing homes for Twitter castoffs. Threads has a lot of influencers and media folks. A lot of big names. Bluesky is full of artists, creators, comedians, and the like. Some of these folks post on both because reach matters, but if you look at where they’re leaving comments it often is Bluesky. Twitter mainstay Darth is on Bluesky and has famously said he’s Never Threads (he is doing his annual hibernation right now, but spring training is coming so he’ll return soon). And Bluesky officially arrived in my view when Dril, the most famous shitposter in the history of social networks, planted a flag there and is doing what, well, Dril does. …

And as an answer to what I wrote for The Los Angeles Times over a year ago, the influx of creators is what I think is the most heartening sign that Bluesky has staying power. Twitter at its best was a place to find interesting people and spread the work they were doing. It was a powerful way around professional media’s gatekeeper effects. Twitter’s death spiral fractured the relationships that made those emerging network effects possible, and where I see them being rebuilt the most is Bluesky.

As an interlude, I’d recommend this interview with Bluesky’s CEO Jay Graber on a recent Hard Fork podcast episode (it starts at 6:06, though the previous three minutes have some good pre-explainer). In particular, she makes a strong case for why decentralized networks are so important.

So all that said, one of the most interesting technical innovations about Bluesky is it offers a different vision for content moderation. Social platforms have been struggling with the idea of moderating content for the past few years. They’re subject to criticism from the public about censorship worries; some of this criticism is quite valid and some of it is political posturing or just made up in some cases. But many of these platforms also have advertisers, and from a brand safety standpoint, advertisers don’t want to have their ads next to toxic or even merely controversial content. So there are financial pressures to do something about content that has extreme rhetoric, obscene material, child sexual abuse material (CSAM), and trolling campaigns that cause real-world harm. There also is at play a sense of ethical duty, something these companies consider to varying degrees — even if the material isn’t illegal or subject to advertiser pressure, some types of content conflict with the company’s vision for how it wants to operate with regards to building a better world for everyone.

On top of that, content moderation is both expensive and impossible. Expensive because it requires a lot of human hands who have to bear witness to some of the worst published material humanity can muster. I cannot recommend enough “The Trauma Floor,” an investigative piece published by The Verge a few years ago. Not because of the particular insight about Facebook per se, but because it describes the life these people live and how harrowing it is because similar to law enforcement needs, somebody has to do this job. When someone criticizes moderation decisions by a platform, I first ask them if they know anything about those jobs and what those folks actually do behind the scenes. Complainers usually don’t. It is hard work because you have to make quick decisions about harmful content that affects real humans and know you’re not getting it right all the time, all while dealing with mental health side effects. Nobody asks for this job. It’s possibly the worst work in online media. The debate over moderation is too philosophical and ideologically motivated, and not grounded in the grim reality of moderation work. And in this work, some content that probably should stay up gets banned because you’d rather err on the side of caution.

Elon famously has gutted the Trust And Safety team at Twitter since the sale, with predictable effects that come with relying too much on AI with a dash of hope-nobody-notices. The most famous recent example is the network’s flat-footed response to the flood of Taylor Swift AI porn imagery being shared on the network (don’t worry, that’s a safe link), something that got so out of hand that the platform had to temporarily ban a search for Swift’s name to cripple the ability to find it.

Social networks have been trying to figure out a human-centered way to think about content moderation. AT Protocol’s method is to let people install what they want on their servers and run their network. But on the back end, AT gives people the ability to create tools to alter feeds that block people or even an entire server that is problematic, creating incentives to stop some types of behaviors knowing that the research says trolls in particular die on the vine when they’re deprived the oxygen of public attention. I would consider it similar to blocking a phone number or a spam email address. Even in the days of Old Twitter, users didn’t have great tools to deal with abuse or disturbing content. Built into AT Protocol is the capability for the community to build widgets and plug-ins that imagine new solutions. While Bluesky the app will have its own rules and ways of moderating, decentralization means it can’t dictate terms to other servers.

But that means other people can use AT Protocol for ill. This is where the big bet is. Right now you can use WordPress or email for ill too. Bluesky the software just creates the ability to communicate, and the critical road test for this whole apparatus is what will happen on individual servers that use the software. Some will have higher standards than others, and so how the ability to block or sever ties with a server goes at the kind of scale possible with AT Protocol is an open question.

I am cautiously optimistic about this. We don’t know how AT Protocol will hold up at large scale. Particular services using AT will get things wrong, and Bluesky the app has not been immune to this (they caught flack last summer, for example, after people discovered holes in the onboarding process that let users sign up with racist usernames and then the company was too slow to respond). This is a giant experiment in that regard. However, I tend to favor social network reform that empowers users with tools to control the experience and beat back content from harmful corners of the internet and creates pathways for the community to speak up about its norms and values, and I expect the types of adjustments and learning that will make these tools better with time. At least someone is trying to do something different when the old system was creaking under the weight of the misaligned economic incentives of online media that only encourage bad actors.

If you’re a Twitter castoff, Bluesky is worth giving a try. You can sign up and immediately do two things to build a network quickly. First, search for people you liked on the Birdsite, and add about 15-20. Then click on that magnifying glass search icon in the left menu. The suggestions it makes are from my experience quite good because it learns from your follows and makes choices that correlate with what you’ve already opted into.

I keep saying that Twitter’s death knell will come when Sports Twitter finds a new common home. I am seeing more sports journalists and news people in general popping up on Bluesky daily. A critical mass is coming somewhere, and in a decentralized world, it doesn’t matter where. All it means is Twitter is in trouble when that day comes.

Jeremy Littau is an associate professor of journalism and communication at Lehigh University. Find him on Bluesky (the app)!