Plus, I already know for a fact it works that way. Because I experienced it. I blocked someone and then later, noticed they replied to me because I had not logged in yet.

Like, I don't think I should have to prove the way the published documentation says activitypub works. This is objective fact of how it works. There's no way for me to know who blocks me unless I admin the server where that person actually is and I modify the code to view it. But the activitypub protocol doesn't publish that and I totally understand why. It's like how Lemmy doesn't show aggregated voting, only the voting of that instance. It's extra info that needs to be added. Now imagine if every blocking action was also now encoded in an activitypub and every instance that read it had to keep that info. Databases would grow much faster than they do now. It's simply not effective. And it'd have to be repeated so new instances also will get it. So you're basically adding at least a daily or weekly posting, unencrypted of who everyone is blocking. All you gotta do is setup an instance and just ignore that data. But then you could easily target people who target you. Being entirely transparent is part of the reason blocking can't work.