I see the problem that in your system the most effective way to earn is just to stake and to go away. It doesn't matter if posts are written or not.
Currently, when self-voting of spam is balanced by down-votes, the rewards of authors and stakeholders will go down when nobody posts. so Stake-holder (curators) need to animate authors to write, so that they have sufficient posts to vote on.
I agree the system is complicated and it needs active downvoting of accounts that try to cheat proof of brain. Also the fact that I need to vote within the first 5 minutes for optimal curation rewards is not optimal.
There must be something in place to animate stake holder for reading and for rating content. Otherwise they will lose interest and go away, knowing they earn interests anyway.
I'm also seeing no need to store votes and posts data directly into the blockchain when there is no influence to the reward.
As a developer, I'm seeing also the problem that your idea needs a major code changes which needs to be applied in one hardfork.
Exactly. There is no reason to turn Hive into just another PoS shitcoin with no reward pool.
Contrary to what @therealwolf or some others may think, Hive (and Steem) have done spectacularly well from a monetary point of view. Pray tell, what other platform that has about 10,000 accounts transacting per day is worth 50 million dollars? For Hive (or Steem) to be in the top 20, we should have a market cap roughly an order of magnitude larger. Not going to happen unless our user base grows significantly.
How to grow our use base? Improved content discovery would be great. Content discovery is a front end issue. Let front ends deal with it. Any developer can fork the Steemit code base - or develop one from scratch - and tailor content discovery to any audience they wish.
Where I disagree with you is the need to store votes and posts on chain. That's absolutely crucial. Storing them centrally would completely undermine Hive's value proposition which is censorship resistance.