This is entirely the wrong lesson to take from this. Why are we still using a plaintext protocol in this day and age? Why can we not get an E2EE addition to the email protocol with full backwards compatibility?
Yes, I understand that it would be imperfect since inevitably not all servers would support it thus forcing additional understanding and decisions on the end user. No, I don't care that a user other than myself might leak my messages in plaintext. Perfectionism in this regard only serves to further shoot us in the foot. Yes, I understand that key distribution is a difficult problem but then that's the case no matter the protocol. Other protocols have solutions that work reasonably well at this point.
There's no justification for the current status quo.
Alternatively I'd be fine using matrix for all my PII related needs (healthcare, government, subscription services, etc, etc) but somehow I don't see that happening any time soon.
Getting from here to there is going to be tough, but I agree 100%. Not only should email be E2EE, but it should include a certificate scheme such that you know the person purporting to be the sender is actually the sender.
How about the metadata? Perhaps if you mean something like self-hosted Matrix, then I agree.
For a public institution you want some sort of accountability / auditing mechanism, so you can't just do E2EE encryption between users.
Otherwise, a public servant could do sketchy stuff behind the public's back with no paper trace.
What you don't want is hostile foreign capitalists leaking your data to their local authoritarians. They are not your public and shouldn't have the data in the first place.
For large organization data the keys would need to be stored within the organization, not with one particular user as in the case of your personal PII needs.
And then you'd still need to worry about digital sovereignity for the keys.