This is a wonderful discussion and much needed. Here are my two cents
The reward curve is one lever and is much discussed about tackling voting behaviour, I don't know if n2 or n or any other shape is better we had problems with n2 but it may help if people discussed alternative solutions more. Your second picture says it all.
The issues we are trying to solve is enrichment of content on Steemit and fairer rewards for posts. There are ways we can tackle this within the current system parameters and maybe we need to try some more initiatives.
There has to be some incentive for voting for other people.
How about a voting bot that rewards people who spread their vote. For every 100 votes you get a big upvote from a whale account or you get on a curation trail from whale accounts? It could be fairly easily calibrated to detect spam voting accounts or automated accounts. Whales could delegate to it.
The more we can encourage manual voting also the better.
We could make a bot that rewards people who manually vote in a similar way by adding them to a whale curation trail.
They are just two half baked ideas but we really get lost in the weeds by spending so much energy talking about the system and the incentives. I would love to change the conversation.
To quote traf, now we have an economy that is paying people $1000 everytime they take a dump on the streets while we have a mayor that thinks building skyscrapers to attract tourists is the answer. Nothing wrong with skyscrapers - it just takes time. Plus the tourists are and will be arriving at a smelly street filled with dump all over, which is why Steemit's web ranking has been slipping away.
If we think about the situation mathematically, n^2 is basically a "tax" on the long-tail under linear rewards, which gets transferred into the top end thereby creating a clear consensus and contrast based on stake-weighted voting. That's basically sacrificing many individuals for the collective. This is the reason why we thought to try something out between linear and n^2 to balance things out. There are a bit of some other tweaks proposed by @trafalgar (check the comment).
Some may say we shouldn't change the fundamental design (of rewards) and add better "offchain" solutions to minimise the trolley problem, including the sentiment that there are no magical numbers that can fix the problem. I think there is a magical region, just like how our planet is in the Goldilocke's zone. Maybe Steem is flying too close to the sun at the moment lol.