However centralization is still a theoretical threat, just like it happened with E-Mail or the various services that started as implementations of XMPP.
And an oligopoly is still preferable to a monopoly. It may be unrealistic to run my own email server, but I can use GMail on my domain and if I want to switch to Fastmail, I can do that. If that’s where fediverse ends up, it will still be preferable to what we have today.
How do they deal with Spam? Domains are cheap, someone could just stand up new instances, federate, dump a load of Spam on everyone and not care about their instance being defederated.
Personally, I wouldn't see it because my main timeline is "following only" (I also have a "followers + replies they make" view for finding new content.) Also I'd be able to remove it because I control my server. Also also I could write a Pleroma filter which says "if it's a post to me and I've not seen that instance before, hide it".
As an instance admin since 2017: we don't have a way to fight this, and it just hasn't been an issue yet because spammers are incompetent. They always just sign up with accounts on legit instances.
When competent spammers start looking at Mastodon, either we hope buying domains is too expensive to be worth it (unlike email spam), or rely on IP range bans.
For one, you can't really spam hashtags since there's no global search or timeline. It only shows up for users if someone on their instance follows the spamming account.
Reply spam still works, but that doesn't scale as easily.
