It gets weird fast, because before privacy controls in the Lemmy source code mean anything, we need trusted third party verification of a server’s patch level, and security controls.
That can be done, and I think Lemmy has a shot at getting to that point, but it’ll be awhile.
In the meantime, I suspect the Lemmy developers are hesitant to add and advertise features that you can’t be sure are actually correctly enabled on your instance.
But yeah, let’s not let perfect be the enemy of moving toward better.
Edit: Assuming you completely trust your instance admin, we could start adding some basic privacy to actions taken on your home instance.
But as soon as the user starts interacting via federation, all bets are off - because the federated instance may he malicious.
I think we might see one or more “trusted fediverse” groups emerge in the next few years, with instance admins making commitments to security controls, moderation, code of conduct, etc.
So, in theory, the lemmy software could start implementing privacy controls that allow users to limit their visibility to whichever part of the fediverse their instance admin has marked as highly trusted.
But even then, there’s risks from bad actors on highly trusted instances that still allow open signups.
Anyway, I totally agree with you. It’s just a genuinely complex problem.