The existing curation rules do not incentivise good curation:

• They encourage users to vote without reading articles. Because the initial curation period is very short. So they actually penalise people who take the time to curate manually.
• They encourage "safe-voting": votes on authors that have previously earned good rewards, irrespective of the actual content of the posts. So we get a lot of posts that are "good enough" rather than great.
• They encourage auto-voting. Hence @blocktrades original post.
• They discourage users from voting on great posts that have previously been discovered - because the available curation rewards are lower. So we get few posts going viral. And as such there is less incentive to produce really great posts.

The aim (or at least the problem I am trying to solve) is how to get great posts recognised and rewarded, to move us into a virtuous circle where people are really incentivised to produce great posts. For me this requires:

• Actual manual curation with people reading posts - this probably needs to be performed by (paid) teams of manual curators (although other more decentralised systems could be possible).
• More time for manual curators to find the best articles.
• No curation penalties for whales / curation teams who wait until the best articles are dug out before voting.
• A simple system (a front-end) to allow whales with little time to see the posts "recommended" by manual curators. Or voting-trails / delegation for users that trust their curation teams.

"If you get 50% anyway, upvoting any random shit is just fine. We currently have 99% upvotes and 1% downvotes per day...If there is no benefit of good curation, why would people do good curation?"

True. People would have the freedom to upvote exactly what they wanted, when they wanted. This is the system that (kind-of!) works for the rest of social media. Given complete freedom of choice, I think the majority of users will vote for great posts. Because, why not? Good for Hive, good for the Hive price if the system really works.

Of course, there are some users who will just vote for their friends. Or circle-vote. But this problem will always exist as long as there are author rewards. And that is more of a problem under the current curation system, since strong voting-circles also attract votes from users chasing curation rewards.

And no doubt some users will just vote on any ten random posts. But, as you allude to, there are downvotes if that situation becomes significant. So there's the risk of those users losing their curation rewards.

"Also, your proposed centralization of power worries me. I don't want like three curation giants to be the main deciding parties of what is good content and what goes to trending..."

Agreed. This aspect worries me a little too. I think to really work the system would need:
(a) A fair number of manual curation groups rather than just a few big ones. Probably specialised by area to avoid workload issues (e.g. one for fiction posts, one for crypto and trading, one for posts with great engagement, or country-specific).
(b) A front-end system for orcas / whales who want to make their own choices but want manually curated suggestions. Making it much easier for them to manually curate.
(c) Some collaboration between curation-groups and whales to all vote occasionally on really outstanding posts. I really think this would be beneficial to encourage quality and for marketing purposes.

In the end, the system I propose does not force any users to vote in a certain way. So users that are worried about centralisation and do not want to support a curation group system can vote as they prefer.

But hopefully a system with more time to vote and some assistance in showcasing great posts will allow users to make better, or at least more informed, choices.