As I expected, my latest article about flagging received a lot of attention. Because there were so many comments, and because writing about a problem without even hinting at a solution qualifies as "whining" - which is something I try to avoid as much as I can - I will also present a potential solution. Since there were already so many proposals, I decided to call mine YAVAP, which, of course, stands for Yet Another Voting Algorithm Protocol.
The Problem
Voting in the Steemit ecosystem can be done in two "directions": up or down.
An upvote dislocates a certain portion from the daily reward pool (around 43000 STEEM generated by inflation) and allocates it to the post for which the vote was cast. The amount is proportional to the upvoter SP and to the vote power. The upvoter also receives curation rewards for his vote.
A downvote will take out from whatever a post already has and reallocate it to the reward pool. The amount taken out is also proportional to the downvoter SP and to the vote power. The downvoter doesn't receives curation rewards and it doesn't get slashed.
The "problem" is made of a few smaller problems.
There is no real Proof of Stake, but rather a Proof of Influence. The amount of Steem allocated by an upvote is not backed by any real collateral, it's just calculated by whatever the upvoter presents to the system as having in his SP account. In other words, there is no skin in the game.
The downvote negates a stream of previous transactions between the upvoters and the post, without getting any penalty. Also, downvoting has no real PoS, because the amount of Steem returned to the reward pool (and taken from the post rewards) is calculated by whatever the downvoter presents to the system as having in his SP account. There is no skin in the game.
Because of 1) and 2) the entropy of the system is high and the predictability is low. No participant knows exactly the value of a bilateral contract between his "proposal" (the post) and the potential validators (the upvoters) hence the state of the system is unstable. Stability (or payout consensus) is reached only after a certain amount of time passes (right now 24 hours, after HF 17, a week). That makes the entire economical ecosystem built around Steem unstable and unappealing for new members. Eventually, the big stakeholders will control the game completely, no new participants will join, the perceived value of Steem will continue to go down (it lost 50% only during the last 2 months) until it will go below the threshold of the electricity cost needed to run the system (just like in Bitcoin these days).
The Solution
Putting skin in the game, for both directions of the vote: up and down. In other words, making Proof of Stake what is is: Proof of Stake.
1. Each upvote will contain a tip equal to the amount of Steem allocated from the reward pool.
Ex: if I upvote and my vote is worth 10 cents, the post will receive 20 cents: 10 from the reward pool, 10 from my liquid Steem account. For each vote, I will pay something from my own account and put skin in the game (or, in other words, I will make use of my stake).
2. Each downvote will incur a CBP (contract breaching penalty) from the downvoter, equal to the amount of Steem taken from the post.
Ex: if a downvote takes out 10 cents from what a post already received, the downvoter will also have to pay 10 cents in Steem from his liquid Steem account. The total amount will be redistributed as follows: 50% will be sent to the upvoters accounts, as a form of compensation (remember, they also paid with liquid Steem from their own accounts) and 50% back to the reward pool (just as it is right now). The amount sent to upvoters will be distributed equally among all upvoters, regardless of their contribution.
Benefits
The reward pool will double in size: for each Steem allocated from there, an upvoter will tip another Steem from his own account.
The upvote and downvote will become more conscious, which will result in a better curation activity, which will result in better content, which will attract new users.
Bots makers will have to work out their AI and make the bots way better than today. A partially automated curation system, which will increase the quality of content is not at all unlikely, but in this setup, because of the costs incurred by an upvote, bots will have to be more carefully constructed.
At the system level, each downvote will send back to the upvoters some Steem, as a compensation for their earlier votes. Psychologically, that will translate in a potentially more rewarding activity, even if the current post will be hit. It creates a positive loop and it still keeps the downvoters' right to not approve content they don't like. The author also knows that the voters received a compensation for their support.
Although the entropy will not disappear completely, it will be much lower. If there is a downvote, the upvoters will be compensated for their contribution by the downvoter.
The demand for liquid Steem increases, because now liquid Steem is needed for upvoting.
Relationships formed will be stronger, although a bit slower in the beginning: if people pay from their own pocket, they will be ten times more careful. That's not a fact, by the way, it's my own perception.
The psychological impact of a downvote will be less significant, since the amount taken out is only the amount received from the reward pool, not the one received via direct transactions (the tip part of the upvote) from the upvoters. So an author can lose maximum 50% of the revenue from a post. Please correct me if I'm wrong.
Proof of Stake will become more than just a mechanism of signing transactions, and it may form the basis for a future governance system. In my opinion, Proof of Stake is not about showing how much SP you have (earned by writing or bought) and using it to manipulate the reward pool with no slashing and only benefits, but about how much SP you can contribute to the platform by curating and promoting valuable content.
Variations
There are a few other variations that may be interesting:
- instead of paying with liquid Steem, voters will pay directly with SP. In this case, the demand for SP will increase
- for each upvote, the amount of tips could be modifiable (with a multiplier, not an absolute value, as it may not be known exactly at the moment of vote), if an upvoter wants to give more to a certain post.
As I know the code now, I don't see any potential technical challenges to this system, but I'm not fully qualified for that, so a technical opinion on that is necessary.
Would love to know your comments, criticism and suggestions.
I'm a serial entrepreneur, blogger and ultrarunner. You can find me mainly on my blog at Dragos Roua where I write about productivity, business, relationships and running. Here on Steemit you may stay updated by following me 
You can also vote for me as a Steemit witness here:
https://steemit.com/~witnesses