Hmm, interesting. I guess it would in some ways be "worth it" if certain users really are a abusive and if there's users willing to sacrifice/park some of their HP to negate others. I like that there's a cost so others wouldn't just randomly negate other people's stake no matter if they're doing good or bad with it but there's also the question of those who don't care about the stake or curation returns, looking at some of xeldal's voting pattern lately and for a long time, for instance, he's been steady under 2-3% APR losing out on a lot of curation returns than if he'd just give up on his current ways and vote similarly to rancho instead on posts already doing well that others won't be able to effect their returns as much. It's almost like he's okay with it as long as it paints Marky/the platform in a bad picture.
Another idea I've been thinking about is if there'd be a downvote pool, or take half of the current one. I.e. Alice votes on Carol and Bob downvotes Alice's upvote but Carol still gets the author rewards that Alice sent her. This would be pretty limited (12.5% mana) so if certain accounts are really misbehaving/have misbehaved badly over a long period of time, there'd be more accounts able to negate their curation rewards without affecting authors. The thing about this idea, though, is that it opens up a lot more abuse vectors and accounts just trying to maximize curation rewards by utilizing their 12.5% mana to take from other's curation rewards even if unwarranted. Maybe instead of the "free" downvote pool it would cost upvote mana instead, so you'd still have to give up some curation returns but the rest of stakeholders got more instead by nulling the curation returns of the "bad behaving account". If their behavior changes the downvoter would naturally stop and go back to receiving more curation rewards while allowing the bad one turned good to earn curation rewards again. Dunno, there was a lot of talks in @dreamsteem's post so was just one idea I was experimenting with but too many loose ends.
The thing about stake-vote-negation is that it is pretty hard to find something simpler and more elegant. It works both to counter curation-related abuse and author-reward related abuse. It greatly reduces the mental overhead involved in policing abuse and it is not technically complex (it's basically just an inverse of delegation).
Stake-vote-negation flips this on its head. Now Bob doesn't get seen as a nice guy just giving out upvotes randomly, but as just some neutered account. Alice doesn't come out as the bad guy.
The main problem with it in the Steem days no longer exists - that is that steemit being a mega-whale could have potentially negated every other player in the system and still had stake to spare. The stake has been much more widely distributed since and there is no longer a single whale account with that level of power that could potentially be abused.
It turns out, Carol was me the whole time.
I hadn't heard of negative delegation before. I can see some merit in it. Delegation is one of the secret weapons of Hive anyway. Any of the tools can be used for evil, but then the platform should be neutral and let people decide how it plays out.
BTW, it is @dreemsteem. You have a typo there.