If the instance admin becomes abusive, or treats the server like his own petty little fiefdom, what can be done about it? Why should I trust an instance admin?

Hi, I run an instance.

If I start acting like a jerk, them my users will start leaving. Pretty soon I don't have anyone left except people who like being treated like that. In my experience that will probably mean they're jerks too.

And thus pretty soon my instance is "that instance run by that jerk who lets her users be jerks". At that point, admins on other instances stop federating with my instance. Eventually the only other instances who are willing to talk to mine are also run by and full of jerks, and I am amidst a hell of my own creation.

So that is, as I see it, the external forces acting on me to try and be a chill admin who's nice to her users. It's a lot easier for them to leave my server and stay in contact with people elsewhere than it is to leave Twitter.

I thought we were comparing Mastodon to Twitter where what you say is already the case

Moving instances is not a big deal and yes, know your admin.

One of the ways in which admins often treat instances like their own petty fiefdom is cutting off communication with other instances that don't toe the line. So sure, you can move instances - just don't rely on being able to contact your friends anymore if you do. (Some instances even have a whitelist of instances that are allowed to follow and interact with their users, with everyone else blocked.)

It's rare to outright block an instance (for not moderating child porn or something of course but not otherwise really)

Instead most "blocks" are just muting from the federated timeline, precisely so that you're not cut off from contacts.

How do my followers know to follow me to the new instance? How do they know it’s actually me?

You can mark an account as having moved and specify to where.

Not reliably, and unless your followers are watching for it, they might miss it.

In good old times we had IRC, where you could go to channel op and report an abusive message. Or to bot master, or to server admin. Depending on personality of that guy (usually they were guys, not girls), it could end with a ban for the offender or personal attack on you. The problem with actual human is that actually you cannot trust him to govern without being accountable, even if you know him and even if currently you are friends. When I checked several large Mastodon instances (>1000 users), I was not able to find any signs of a formal and legally binding policy, which means that their admins are not accountable and you have no more rights than a slave of a pharaoh. Mastodon instance may protect you, but is not required to do so and there's no guarantee that it will not support your offender or it will not suddenly abuse you.

> I was not able to find any signs of a formal and legally binding policy

Every single centralized service (Hacker News included) reserves the right to ban anyone for any reason. Twitter may have a policy, but in practice it's no different than anyone else's Code of Conduct. It's certainly not legally binding.

Regular users don't care about whether someone could theoretically become abusive in the future - maybe they should care, but they don't. They just look at the track record and decide if they trust that person right now.

And right now, Mastadon instances are better at moderation than Twitter. You're worried about an instance admin theoretically not protecting you in the future; Twitter already, today, doesn't protect you.

That's exactly the point: regular users don't care, so they won't consider Mastodon based on whether it can be trusted or not. For those who care Mastodon pretends to provide a trustworthy service, but in fact it's no better than Twitter or anything else. Even if current situation with moderation is better, it's the same as to advertise participating in some deadly stunt on the ground that "noone died yet". Where are the guarantees that an instance admin that is "out of patience" (you can find such example somewhere in comments here) would not decide to ban you from her instance in a bad mood?

But in practice small instances are better at this than large platforms, and in practice flexible moderation policies are pretty observably better at curbing abuse than ridged ones are. The point isn't that Mastadon is perfect, it's that it's better. In business, better is good enough.

A lot of people are OK with a middle ground where moderators remain, but are closely associated with their communities to reduce communication problems and increase their stake in doing a good job. Is it really crazy to believe that someone might want to trust other people, but also want to have more selection over who they trust?

You're here, so I would hazard a guess that you're OK with that middle ground too. HN admins could theoretically turn this site into 4Chan or Reddit if they wanted to. But we both trust them not to, and the fact that they theoretically have the power is not enough to make either of us throw up our hands and say, "well there's no difference, and nobody will care, and there's no reason for anybody post on HN instead of Reddit."

There are a lot of people here posting on a dedicated tech-community Reddit clone about why fringe Twitter is a bad idea.

Mastodon has literally no moderation tools. There is absolutely nothing to prevent impersonation or harassment.

That's incredibly inaccurate. There are a number of different tools available to users, and even more available to instance admins and moderators.

Source: I'm a moderator on anticapitalist.party. AMA.