Hi @teamsteem, I think that it is good that you raise N^2 vs linear debate again. The survey conducted by @cervantes indicates that the top witnesses are against N^1.3 (mild super-linear) but it wasn't discussed in much depth in that post.
Before too much is thrown around regarding solutions. The problem/s we need to solve need to be clearly stated otherwise the average Steemium is going to question what all the fuss is about. What would help would be a post clearly stating the problems. Are we talking about a rewards problem? Quality of content problem? User interface problem? Content discovery problem? Flow of information problem? All of the above or some of the above. I believe it is a mixture of all of the above.
People generally try to get the highest rewards for the least amount of work. The low effort options, at the moment, are more rewarding than the higher effort options such as creating content and manual curation. Rewards should align to value added to the community. That is just a very brief summary of what I think the problem/s are. This should be articulated a lot further. Once the problem/s are clearly articulated (i.e. majority of the users can understand the problems). We should discuss possible approaches and solutions.
Mostly likely a combination of solutions are required. 50/50 curation split plus separate pool for downvotes appears quite popular with witnesses and fairly split amongst users. You discuss N^2 rewards as a response to 'selfish behaviour'. All these are Steem solutions. The weaknesses of the UI has not been discussed in combination with Steem solutions. There is a lot that can be done with a user interface. Steempeak has already taken a step forward with the use of saved topics, trails and users.
It would be good to look at as many possible solutions and scenarios and how they would work together to solve problems that have been clearly identified and understood by the majority.
Thank you for your very eloquent and on point contribution.