As Twitter continues to descend farther down the proverbial drain, it seems like storm clouds are forming over the Fediverse as well. Recently, a game was released about a certain fictional school for young wizards and witches. The author is a controversial figure with regard to transgender rights and feminism, among other topics. As such, a number of aggrieved people flooded Facebook, Twitter, and every other medium they could think of with diatribes about the people purchasing and playing a game for children. Many decided the best way to express their dissatisfaction was by posting spoilers for the game, so that others couldn’t enjoy it.
I’ll do my best not step into the controversy itself, because I’m working toward a different topic. The focus of this post is the “splash damage” that the Fediverse took from the event, and the conclusions I drew from it subsequently. Let’s begin with context: Twitter, Facebook, etc. all did their usual, awful job of amplifying the conflict through the use of algorithms that leverage anger to increase “engagement.” But ActivityPub services like Mastodon don’t use an algorithm, so everything stayed calm there.
Just kidding! It was a nightmare in the Fediverse, too.
It appears that mainstream social media’s ability to radicalize people and harness manufactured outrage doesn’t just stop at the borders of those services. Angry, irrational people came surging into every Mastodon instance, demanding that anyone playing the game be sanctioned or banned for… anti-trans and antisemitic behavior? The leaps of logic required to reach those goalposts were breathtaking. Some people went so far as to say that anyone playing the game was participating in “genocide.”
As a reminder, this was all over a children’s game about wizard school.
This is when the wheels really came off the wagon. The biggest example of how badly the Fediverse handled the event was on an instance called mastodon.lol, which housed roughly 18,000 accounts at the time. The mission statement for mastodon.lol is included in this screenshot. Note the priorities that the administrator enumerated in the instance’s description.
Well that seems like an instance likely to survive the controversy, right? It’s an LGBTQ+ server by definition, and it’s run by an administrator that flies the rainbow flag emoji in his name. Surely the outrage would be reserved for more conservative or generalized instances.
If you thought any of that — and you could easily be forgiven for doing so, as they seem common-sense things — you were far from correct. The admin, Nathan, posted a statement that game spoilers weren’t allowed for Harry Potter. In essence, his argument boiled down to “let people enjoy things.” Here was his initial statement:
At first blush, this seems a reasonable policy, and it also addresses the concerns of those who were angry about the game. The administrator empathized with the motivation behind the backlash, but also wanted to protect the rights of those who just wanted to enjoy a video game for children. This statement immediately generated a good deal of “feedback.”
Wow, that escalated quickly, right? Now both the backlash and the backlash to the backlash are well into irrational territory. But that’s surely where this public debacle plateaued, right? I mean, how much more insane could things get?
Hold social media’s beer and watch this.
At this point, all reason and compassion are lost on every side of the argument. One party arguably had the high ground at the start of this disaster, but under pressure succumbed to the temptation to “hit low and hard.” These comments, predictably, threw gasoline onto the metaphorical fire. Other instances began noticing the fracas, and decided to take varying degrees of action themselves.
THIS is the part I’ve been building toward: the part where the tsunami reached the shores of meow.social, where I had my primary Mastodon account. Now a responsible admin might think “This is an instance of 18,000 people with a single, overwrought administrator under fire. Perhaps the best course of action would be a temporary (or permanent) mute/ban for that one person? Maybe a statement of principle and policy? Or perhaps a call for cooler heads and a ‘time out’?”
Nope. One of the admins at meow.social decided the best approach was to defederate an entire instance, catching 17,999 other people in the crossfire over one person. The other admins stepped in shortly thereafter and pointed out how incredibly over-the-top that reaction was, and how it hurt people who were trying to leave the instance and migrate to meow.social. It also pulled apart follow lists on both sides of the suspensions, meaning that people on meow.social suddenly lost all their mastodon.lol followers and vice-versa. When the admins undid the change, they discovered that the ensuing suspensions couldn’t be automatically reversed, and instead pushed that work onto the people harmed by the change.
Sadly, this wasn’t the first time the admins at meow.social had made similarly hairtrigger decisions. An ISP once announced that it would allow its customers to log into their Mastodon server using a Single Sign-On arrangement for their convenience, and the admins at meow.social lost their damn minds. They defederated the entire instance and claimed it was about to flood the Fediverse with thousands of zombie accounts. It took an enormous number of people to explain that no, accounts are not auto-provisioned. They’re provisioned at first logon to the service. So aside from being reactionary, the statement was also completely untrue — more Twitter-style disinformation amplified as fact.
There had been other unusual things going on between those major events, too. Sometimes I would try to browse to another user’s profile on a different instance, and get a warning that the meow.social admins had flagged the server as suspicious. When I clicked through the prompt, I would normally discover something completely benign, like an RPG geek’s feed or someone who did art photography. One time I even got a slap on the wrist while I was working out an integration for my Untappd account, despite the fact that I was cleaning up test posts in under one minute.
The Great Harry Potter Explosion of 2023 motivated me to abandon meow.social. The admins released a half-hearted apology, promising that nothing like this would ever happen again, but to me it was more than three strikes on their part. I felt like they largely governed by caprice and grievance, not fact and reason. That’s when I set up my own instance: ohbear.wtf. (Finding a short, punchy domain name is super tough these days, but that one always makes me smile and snicker.) And that is when I had a revelation: there isn’t a computer algorithm on Mastodon, but there’s sure as heck an organic algorithm.
What do I mean by that? Allow me to clarify. Look at these side-by-side Federated Mastodon feeds, one from meow.social, and the other from ohbear.wtf:
The content in the two feeds is strikingly different. The one from meow.social seems to emphasize the usual outrage that circulates at Twitter and Facebook, but the one from ohbear.wtf is mostly comprised of personal information and creative projects. If both of these instances federate with the rest of the ActivityPub world — including each other — why is there such variation in content?
For the answer to this question, we need look no further than Qui-Gon Jinn:
Our focus determines our reality in many different ways, but one of the major ways that applies to modern technology is how Mastodon servers federate with each other. In essence, your instance “learns” about other instances based on what people choose to follow. On meow.social, which has thousands of users, “outrage posts” dominate because so many people subscribe to them. On ohbear.wtf, which has a whopping two users, those posts are absent because the two people — myself and someone I trust a great deal — don’t choose to focus on that. Ergo, the instance never discovers the lion’s share of that outrage.
This brings up a number of serious concerns. Can people un-learn what the algorithms have taught them? Must anger always be the most compelling content on the Internet? Can the Fediverse survive in an environment where instances defederate each other over the slightest provocation (like a video game for children)? I can’t answer those questions, and as a pathological pessimist, I probably shouldn’t try.
As for mastodon.lol, the damage is well and truly done. The administrator has had it with running an instance with thousands of screaming, hostile voices, both locally and on other servers. Additionally, his angry comments have demolished user confidence in his ability to run an instance with thousands of screaming voices. Even if he changed his mind, the instance would likely become a ghost town. It’s already lost approximately 1,000 users.
So in May of 2023, the instance will shut down altogether, leaving thousands of people to either scramble for a new one or leave the Fediverse entirely.
Western Civilization makes better computers each and every year, but it never seems to make better users. Ultimately, that limitation could spell the early end of the Fediverse. An old axiom from Computer Science definitely applies here: garbage in, garbage out. And social media is largely garbage these days. The absence of algorithmic content promotion only helps if people let it help.
