Several days ago, an insightful article was posted by 
Cited from the mentioned post:
We currently have a system where stakeholders are incentivized to engage in content indifferent voting behavior (vote selling/self voting) to maximize returns. This completely defeats the protocol’s ability to foster engaging platforms that revolve around content discovery.
We need to move away from an economic system that makes it too costly to vote on work that stakeholders actually deem valuable if we want this place to survive:
- Upping curation rewards is an obvious answer, but by itself, it is easily circumvented. Bid bots can just offer to share most of the curation rewards. Self voters can spam hundreds of posts and just upvote the ones that were least voted on by others after 15 minutes to avoid others 'stealing' their rewards.
- Non linear rewards curve should be seriously considered. Having reward amounts/SP that differ per post depending on its popularity is the core component of moving away from content indifferent voting behavior. Yes it has a cost, but the n^2 we were use to was far too steep. At n^1.3 (or any crude approximation of this), we should get most of the benefits at the lowest cost.
- Downvotes incentives should also be encouraged. Of course greater downvote incentives has downsides, but in combination with 1 and 2 above it should allow for curation rewards to out compete mindless vote selling/self voting.
Content indifferent behavior is the bane of this system. No social media platform can survive if it's unable to at least sort exposure and rewards by some subjective standard of quality.
Based on this ideas - which are more deeply explained in the post, a set of concrete implementable changes have been proposed and discussed in public and private channels in a rather unordered way.
To facilitate the discussion, involve further community members and prove the viability of the modifications, the majority of top witnesses have all agree to individually show their willingness to support the modifications by first a) Identifying a possible set of changes and b) Individually vote for its approval or disapproval.
Identified changes
This is a the list of change proposals:
n^1 rewards curve
The rewards curve shall be kept lineal as introduced in HF19 (keeping it as it is today)
n^1.3 reward curve.
The rewards curve shall be updated from lineal to a “mid-superlineal” n^1.3
50% Curation reward.
The reward distribution per post should be updated from 75% for authors and 25% for curators to 50% and 50% respectively.
10% free downvotes
Users shall be able to downvote 1 post/day at -100% without actually decreasing the amount of remaining voting power
Separated upvote/downvote pools.
Users shall be able to downvote 10 posts/day at -100% without actually decreasing the amount of remaining voting power but consuming its separated “downvoting power” from its own pool.
5 minutes curation window.
The curation window shall be reduced from 15 minutes to 5.
3 minutes curation window.
The curation window shall be reduced from 15 minutes to 3
Consensus status
Witnesses have been asked to approve, disapprove or just accept the changes to summarize the actual consensus status. Meaning:
APPROVE: Witness is actively for the change
DISAPPROVE: Witness is against the change
ACCEPTABLE: Witness wouldn't vote against (if majority does) but doesn't necessarily approve of it
EMPTY: Witness have not express any opinion yet
These are the first results of the consensus voting:
Explanatory comments by some witnesses:
:
0/100 Experimenting this change with SMT is best way for now, 50/50 I strongly believe this will improve engagement
I'm still of the believe that a 50/50 change won't serve the average user and will only benefit the largest stakeholders and bots. The curation rewards system itself would benefit more from a complete reinvisioning than just turning the dials. If the majority of witnesses want 50/50, I won't stand in the way, I just don't believe in it.
Downvote should be 50-80% of upvote, also do nothing till SMTs come out
On superlinear (n^1.3):
We already have a small degree of (something like superlinear) at the low end. The dust vote threshold is subtracted from every vote, so for example a vote of twice the dust threshold only has 50% weight relative to SP. This converges to linear as the vote becomes significantly larger than dust. I would be aggreable with this being modestly enhanced. What I don't want to see is superlinear at the high end.
On curation window:
I'm in favor of the shortest proposed window (3m in this case) as I actually think the window should be even shorter, like 30s-1m. I don't think it should be used to force humans and bots into head-to-head compeition. Instead I see the value of the window as a short opportuntiy for bots to bid down the curation rewards on easy sure thing type content (known authors etc) for which curation is low value. These rewards forfeited by bots via this auction rewards are returned to the pool and can be paid out some else where they have more value. Once the short window is over, humans should not ever need to 'time' their votes.
I left comments here
Curation: would be open to a sliding scale adjustable by author
50/50 is still too big of a change, propose a test at 60/40, bad psychological UX for users to lose half post rewards, hinders business models of current major apps utopian, fundition etc, curation will still be centralised around certain circles, who can produce good content to avoid downvotes. Careful incentivising downvotes, could literally lead to a downward social spiral and no longer the "Dinner party" civil mentality ned once called for.
I approve of a very mild superlinear reward curve but only if it can be implemented without a large computational penalty to Steem nodes. A curation penalty window of 10 minutes makes more intuitive sense, but like a 15 minute one, a 5 or 3 minute window doesn't. This is an issue that can help or hinder voting cognitive load. A scale of 10 minutes easily translates to 0-100% curation burn.
I would like even more rewards for the curators if possible
Overall I don't think tweaking the reward curve or curation % is worth focusing our time and efforts on. The vast majority of users don't understand any of this stuff and I don't feel like changing it will change their behavior or encourage more people to invest. I do support the separate downvote pool however.
n^1.3: I'm in favour of testing out the n^1.3, preferably on a testnet which resembles the mainnet | Downvotes: I'm in#terested to see how this turns out, but still mindful about the consequences this would bring to toxicity.
I am in favor of a witness controlled downvoting pool size. So we can roll it out slowly with low free downvotes at first and see how the community responds and then adjust the size depending on the results. If it's hardcoded we won't be flexible enough to react.
The table above might contain errors and need to be seen as a work in progress that, if required, will be adapted as soon as witness express further opinions and/or vote rationals.
As already mentioned, we invite with this post all interested stakeholders and community members to further discuss the topics, express concerns or amend errors.
EDITED (26.10.2018): The table has been updated with the latest information and the individual comments by the witness added.
EDITED (27.10.2018): 