I keep seeing this analogy and unfortunately that's not how email servers work so it never really helps honestly. The servers are the To: fields, not the From: fields.

Okay, I know that the sender of a mail can be faked to a certain degree, but if stuff is setup correctly on both ends, you can verify that an email actually is from where it is saying it is.

Even if anyone could use any email-address to send from, the point still kind of is the same: You don't have one single mailserver, where the people are required to be on that server in order to message other people on that server, but you can send messages from a different server to that target-server, where the user is residing on.

And there's also no real analogy about privacy. With most email providers the intent isn't that everyone reads everyone else's email.

This is true, but it isn't the point either with the example

So frankly I really don't know what insight this is supposed to provide if it doesn't behave like email.

The point is that with Twitter, Bluesky, Facebook, Instagram, every one of those platforms is closed to the outside (even tho I think Bluesky is or was thinking about opening to ActivityPub?), you create an account on Facebook and you can use Facebook, message everyone on Facebook.

With Mail, if you want to write Mail, you need a mail account from any provider, like Google (Gmail), Microsoft (hotmail?), or can host it yourself on a server of yours. Then you can write a mail to anyone who also has a mail account (which your server hasn't blocked and whose server hasn't blocked your server, which happens for example when your server is misconfigured and is allowed to send malicious mails).

It's the same with Mastodon/ActivityPub, if you want to message someone on ActivityPub, you need to choose any provider (Mastodon/yada yada/..) or can host a server yourself (which in turn can block other servers and can be blocked by other servers).

Of course there are technical differences and mail usually is 1-to-1 (there exist mailing lists though, which is basically 1-to-many/all), encryption is handled differently, but the key in the argument is that you need to choose one provider out of a list or can host yourself and after that you can message (mostly) anyone on other providers.

And there's a big safety difference. With something like Bluesky you have to trust the server admins to behave. With ActivityPub you have to trust each and every user of the service.

Why do you have to trust every user? Because they can send illegal content? Users can also do that with Bluesky.

Which is why server admins get shirty about whether they will forward messages to or from other servers. That whole situation doesn't really exist with email.

I'm probably not understanding the example you want to make. If you are really talking about the example I made above, as I already said, on any service you can send malicious/illegal content.

It's not like you have create a Hotmail account because Gmail has decided to defederate with Google or whatever.

I wrote about this above. Mailservers can actually be blocked by other mailservers, this happens quite frequently, as written above, when a mailserver is misconfigured or also when a usually "small" mailserver is suddenly sending many mails out, for example because the admin/owner is sending a newsletter to many users or invites to some event or similar.