In the recent few weeks, like millions of people, I jumped on the Mastodon bandwagon. After over a decade on Twitter, I was ready for something else. And Musk’s clown-show gave me the push I needed to do a leap of faith.
I wanted to create an account on the biggest, most “official” looking instance (mastodon.social) [14% of total active users], but too many people had the same idea and registration was (and still is) disabled on it.
I sorted instances by number of active users, and the next 3 were mostly for japanese content [collectively 20% of total active users], so I picked the next suitable one (mastodon.online) [2.6% of active users] which is apparently administered by the same people from mastodon.social (I don’t know why they setup 2 seperate instances).
On Twitter, Mastodon evangelists were telling people “join whatever mastodon server you like, they all talk to each other anyway”.
But this is not true. After a few weeks of usage, I discovered the hashtags #fediblock and #fediadmin and the power dynamics of instance interactions.
The administrators of a Mastodon instance have a number of tools at their disposal to get rid of spam, ban illegal content and enforce community guidelines and rules.These tools can be applied to the admin’s instance, but also to how it federates with other instances.
This can range from muting or silencing some words or hashtags, to blocking particular accounts from interacting with your instance, to fully blocking other instances from interacting with the users of your instance.
Some blanket blocking of some instances is fully justified and is necessary for the health of the ecosystem. No sane person wants their timeline to look like 8chan.
For effective moderation, admins are also sharing lists of blocked servers, so that the worst offenders can be proactively blocked and effectively quarantined from the rest of the network.
Out of curiosity, I wanted to check the reasons provided for these instance blocks. And most of the time, they make a lot of sense (racism, homophobia, etc…) but some listed reasons made me think of the power Mastodon admins are wielding on behalf (and on) their users.
Without any particular order the ones I found funny:
- Dorks and shitlords
- Shitposting
- Fanboyism
- Bootlicker admin
Then those that don’t like people who talk about politics:
- Heavy politics
- Tankies
- Politics without content-warning
- Discussion about chinese politics
And then there were the reasons that made me doubt the federated model of moderation:
- Ineffective moderation
- Moderators did not remove a user
- Federates with instances that violate our terms of service
This basically means that if I join a server, my friends join another. Then admins decide that my server is not good enough because it federates with a server they don’t like for having “heavy politics” then I would lose contact with my friends without any action on the user end.
Admins can also choose to not display the list moderated servers on their instance, so this can be done without transparent logs, and suddenly I’m not following people I want to follow.
I believe that the current model of distributed moderation is flawed. Server admins can decide unilaterally and subjectively what the users on their instances can see. This can be an effective tool to combat spam and hate-speech.
But without transparency, consistency and effective policies this can turn into a small cohort of benevolent dictators that will block you from talking with your friends because your server does not block another server that breaks some other server ToS. So far I’m not convinced by this model of giving this much power to Reddit style moderators, who are trigger happy with full blanket blocking, because they don’t have the time and resources to do more granular moderation. The current model doesn’t scale.
PS: stats and data collected from https://instances.social
Your frustration is warranted, but in practice I fail to see how this is a problem. All distributed services (BitTorrent, I2P, email) grapple with this issue, and Mastodon simply turns it into a feature with a button attached. Even if Mastodon's codebase was redesigned to restrict censorship, admins could still firewall instances they disagree with or fork Mastodon to create an even more discriminate ActivityPub application.
The same thing happens with email, but that hasn't stopped it from scaling. If your message hails from any place that isn't one of the top 5 email providers, most services will mark it as spam or delete it outright. If you want to talk to your friends on a highly-political instance, you'll need to find a homeserver that agrees with you (or start your own instance and bear the weight of moderation).
As a counterpoint, I'd argue that centralized moderation doesn't work either. Trying to ban actually hateful content on Twitter is a losing game. The fact that everyone is exposed to everyone else makes it ripe for harassment, which is something the Fediverse explicitly avoids at a protocol-level. Queer spaces can decide to federate exclusively with other queer spaces, if it makes them comfortable. Conservative and libertarian instances can do the same.
If you're not paying for your own instance, you don't get to decide the rules. It's not "Reddit style" moderation because the moderators and content hosts are the same people. They're not being given any power that they don't directly deserve as the proprietor of their instance. It's understandable if you're frustrated by the differences between Mastodon and Twitter, but making claims like "the current model doesn't scale" is plainly wrong. Mastodon has scaled, it just doesn't have any interest in becoming a sequel to Twitter.