RE: Specialization - jump starting new steemit sub-communities and creating insurmountable obstacles for others
I very much like the idea of sub-categories. Communities would form around common interests. This solves a huge problem of interest based discovery. It also helps with scalability quite a bit.
Here is another potential approach I've been thinking about:
This model would require a change in how the rewards pooling is done, to ensure a fair distribution of rewards between communities . Perhaps each community would have its own rewards pool, proportional to the size of the the community and total SP of its participants.
If a person is a member of multiple communities, their SP influence and rewards are split between all communities that the member participates in, proportional to their SP pool sizes.
For example, if I am member of a gaming community with 10M members, who collectively claims 150M SP and a music community with 5M members which collectively claims 50M SP, and I have 200 SP myself, 150 of my SP will go to the gaming community and 50 SP to the music community.
Human decisions are key, see my response above about registered dolphin curators. I dont think there is too much change needed to the system. Just getting the unused steemit's voting power is all that is needed. the rest can be done via overlay to the existing system
The rewards segregation based on community size (or better yet its cumulative SP) would make the rewards fair, while maintaining the difficulty vs reward ratio equalized across the platform.
Unfortunately, it would also cap the max a post can earn to the communities rewards power.
Perhaps I am thinking too ahead of the time, I see this more applicable to when Steemit has millions of members.
So in the current stage, I think your solution with dolphin curators voting on steemit's SP might be a lot more practical.
I agree that ultimately something more like your idea is good. but one step at a time, minimal changes to the existing system that solves the current obstacle. Iterate and we dominate. Stagnate and ...