Good post, and I definitely agree with most of what you've written.
One thing I'd pick up on is that although communities are generally best placed to moderate their members (and have the specialist knowledge to do so in more complex cases) I think there is still definitely a role for Hivewatchers.
There two scenarios I can see. First because it's possible for a user (at least in Ecency) to post to their own blog rather than a community and then use non-community tags to get the post noticed, thus avoiding scrutiny by any particular community. Second, that there are small communities which are effectively abandoned by their original creators, and could thus be posted in by someone who was deliberately skirting the rules. In both those cases, I can see Hivewatchers being an essential safeguard.
I can also see the Hivewatchers role evolving to be more of an arbitrator role, where users can appeal if they think a community is moderating them due to inter-personal conflict rather than any breach of the rules.
But I think what might be needed in all of these situations is for Hivewatchers to develop some. There should be clarity over exactly what the role is, who is carrying it out, how it is rewarded (and it should be, just like any other job !), and how new members can be added to the team. Also, I feel enforcement action should be graduated according to the circumstances, with an emphasis on being more educational than punitive for first-time offenders or for high-volume posters who might have slipped up rather than deliberately breaking the rules.
I guess what I'm really saying is that it should just be a human thing, not a rigid application of rules that could be done by a machine !