I agree with your logic but without downvotes we have nothing to prevent large scale automated abuse and plagiarism earning.
That's a much worse situation to end up in, unless you can activate so much stake to actively curate that their minimal earnings become all below dust vote levels and the possible scale they could generate doesn't matter anymore.
The biggest problem is humans like to behaviorally suck a middle ground solution needs to be found between satisfying emotions and still having tools to fight genuine abuse.
The EIP curve is a solid start but want to make it even harsher for lower appreciated content (fight automated farms) that's also harsh on not discovered creators who post for $TU 0.5-3.
It's a very complicated topic and many opposing positions can all be right in their analysis and sentiment.
!ENGAGE 35
Do you like private property?
Give "downvote power" exclusively to the witnesses.
That way, we can at least "vote" for the witnesses we agree with (I guess you can't "downvote" witnesses currently, doesn't that sort of break your "must-have-downvotes" moral axiom?)
As for the "rampant plagiarism" "problem", cheetahbot already provides links to the "original content", which sort of automatically "fixes" the "problem". Anybody who forgets to post attribution links will be thankful for cheetahbot's helpful link posts. No downvoting required.
And to fix the zombie-sock-puppet "problem", just ask @steemit to stop automatically delegating 15 steem-power to every newb. This would essentially make steemit "invite only" because you'd either have to buy some steem yourself or be sponsored by an existing account. Don't forget gm.ail and fa.cebuk (and others) were "invite only" when they were getting started, it builds "community" and gives an air of "exclusivity" and "prestige".
And to fix the self-voters, disable self-voting (most websites already disable self-voting automatically).
And to fix the circle-jerks, disable voting on the same account more than once every seven days.
You could set the parameters to only allow a maximum of 25% of your running seven day votes (by total amount of steem) to be granted to the same account.
https://steemit.com/ethics/@randr10/q5zczj
ALSO, LOWERING THE MINIMUM PAYOUT TO 0.001 (would remove the financial incentive to downvote) AND FLATTENING THE CURATION REWARDS (NO MORE INCENTIVE BONUS FOR VOTING IN THE FIRST 5 MINUTES) WOULD GO A LONG WAY TO "FIXING" STEEMIT (are the band-wagon voters really "adding-value"?)
SOURCE CONVO
I see a lot of misconceptions here. But will be brief in reply.
Witness voting and "must-have-downvotes" are two entirely different processes. Yes, you can downvote any comment/post created by a witness. No, witness voting does not have a downvote mechanic. I think one needs not explaining why not in a "delegated proof of stake" system.
You underestimate the degree of automated spam abuse which has required so far resource credits and EIP already. Lowering dustvote rewards payout to 0.001 would be counter productive and only increase the earnings by such actors again.
A certain degree of abuse/maximization will always happen. It can not be fully nuked and that should not become the priority either. A system which is well-designed will minimize the scale of the abuse, with a scaling solution.
Emotional butthurt will always happen over downvotes. That proves a misunderstanding about the rewards belonging to the community until a post is payed out.
Humans bound to maximize will always find ways. Same with circle jerks. The latter are very hard to recognize as patterns without high levels of false positive and thus collateral automated damage. That is why also manual downvotes
Dapps, and future SMTs, can decide their own rewards system.
Sybill attacks (alt accounts) are a reality. If I can dedicate only 25% to an account, then I create 16 accounts if I want to maximize. (Don't get me started about IP recognition aso, maximizers know VPNs)
Also, there is no reward for activity. Unless you're an actifit user.
I understand that downvotes are a fickle thing to understand and accept. They are indeed weird and may seem hostile, poor design even, because entitlement over rewards can easily grow. But, the chain owes nobody anything until the final rewards amount is locked and subsequently paid out.
I too would much prefer that retaliation and similar sentiments were not part of the human modus operandi but they are. That still doesn't mean the system is flawed because many people fail to understand that nobody is owed anything by the chain. It's a weakness in the system because humans.
If you have a system which both understands the scale of (automated/spam) abuse and finding the middle path I will gladly read it.
So far you don't. That mostly because you don't seem to acknowledge the vast amount of maximization/abuse which needs countered as much as possible (because takes away potential rewards of real users).