Of course he has special rules for himself. Fun fact: you cannot block Mark Zuckerberg on Facebook. It just won't let you.
This was fixed months ago (and was a technical problem that affected anyone blocked by many thousands of people, not just him). It wasn't intended.
We have been hearing the "it's a bug" a lot lately from Facebook.
Their backend and frontend code is likely also a huge monster. In addition they're doing ML/AI stuff which is even harder to debug.
I don't imply it, but given how many spam filters these days are realized as more-or-less-black boxes, it is a plausible theory.
In addition, at least for Twitter and FB, there has been suspicion for long time that reports/blocks have faster consequences when many people report/block, as automated filters take over before a human can look at it. I can imagine that there is a filter that gets triggered when a user gets N reports/blocks in M minutes, with e.g. N=100 and M=30, to automatically flag users as potential spammers, unsolicited nude senders or whatever. This had hit Zuckerbergs account, the spam filter blocked his account... and then someone was tasked with preventing this, by disabling the block function based upon certain criteria.
Also, the bug seems to have affected only users blocked by thousands of people. Not exactly your usual testcase, and I don't know if I'd have thought of this scenario when writing the test specifications.
Building products is all about finding the right balances...an engineers mind often wants to expose all possible options to users, just because they can.
But I have never seen that work with users at scale...you need to curate, only give people the controls they absolutely need