Plenty of email servers will throw you off for using the domain to email things they disagree with. It's just that most people have migrated to general purpose providers for their personal addresses.
Modern email most definitely does block domains and distrust unfamiliar ones (I extend my personal guarantee that getting around spam filters will be an issue if you host your own email), and there most definitey are such things as domain restricted email servers.
Moreover social media faces distinct issues that require specific tools, causing the email analogy to break down a bit. A lot of organized harassment campaigns leverage network effects of platforms in ways that wouldn't work over email. The per instance federating capability helps to mitigate that, and if mitigating harassment is not something that is a priority there are instances you can join that take the open-to-all approach.
Those domains are being blocked because of what they send.
They're not being blocked on the basis of what domains they block. There's no recursive blocking that wants you to take a stance and punishes you for being a neutral acceptor of emails. Those blocks are the behavior that can make federation suck.
I accept that I overlooked your distinction here. You are right that there's no recursive blocking with email.
But I also think that's an oversimplification, because the block list is a helpful proxy for content characteristic of an instance.
But this is a case where I think the analogy to email breaks down insofar as its instructive to social media generally, because some things about email don't cleanly map on to what federated instances are managing. It makes sense that you would want to block instances and their enablers, to connect to an ecosystem consistent with the intentions of a particular instance.
As parent commenter said you can use a general purpose provider (with email) or general purpose instance (with Mastodon), so you are accessing the fediverse. The only thing you are "missing out" on is the ability to make it compulsory for other users to listen to you against their consent.
I feel like "people that talk to them" is going a bit far as a definition of "enabler".
And social media posts in this design are pull, not push, so nobody is ever forced to listen to anything. If you block an instance you won't see their posts, no need to block any intermediates.
The idea of the fediverse as an agglomeration of private forums is a half-decent mental model. Each instance has its own rules and community, and some of them really hate each other.
Maybe. You'll frequently find many stores/communities/services saying "hotmail/yahoo/___ users, we haven't been getting your emails" because those services aren't quite as big as Gmail and are getting spam-holed or whatever. I'm sure spam filtering already gives a major hit to free email services. If a mail server started up that explicitly permits in its ToS propagation of hate speech and terrorist threats, it's probably going to get blocklisted by all major email services, no?
You may lose access to your Google account. I run my own mail server and honestly despite always hearing that people have problems when doing this I've had 0 trouble in years.
My email provider doesn't prevent me from sending an email from someone at Gmail because someone at Gmail happens to have opposite views to the admin.
Yet email is still federated.