Yeah, all the difficulty will be in finding a middle ground between optimizing curation rewards and discovering new content.
Don't want to be pessimistic, but from my point of view, they're incompatible.

For example, when OCD discovers new content, it in fact only profits to anomadsoul, which front-runs the votes from ocd, ocd-witness and ocdb with his "triggering" account cameraman.
The only other people making curation rewards from content "discovered" (upvoted) by OCDB are "lucky voters" who casted their votes before, or insiders (I don't know how they work really, but I suppose the person who suggested the post). In fact I suppose most of those "lucky voters" are in fact the "insiders" or at least partly.

Sure, the vote from tipu (around 1$) might give the post some spotlight and trigger votes after it (if it arrives in the hot section for example), but I wouldn't overestimate that effect.

For me you said it all here
> good authors that create valuable content to be rewarded, even if it's done by voting circles

If you want good return on curation rewards, you will need to study the voting circles and vote at the right time, or front-run all the tipu votes with your cardboard account (in that case, you'll become a great performing curator, like anomadsoul is :D )
If you don't care about curation rewards, you can start "discovering content"

As Tipu is a business with delegators behind, that's why I would separate both activities which, are, as I said, in my opinion incompatible :
- An optimized curation system to ensure a good return to delegators,
- An open curation initiative to allow people to curate with higher votes, with high turnover among curators to really allow new authors to be curated / discovered

Hi @cardboard I am a very active curator (cofounder of @c-squared / @c-cubed curation group) and would love to participate in your project as a curator.  Would be great to be able to give an extra boost to those really awesome posts I find and should be a good bost for tipu as well as a ton of big votes often end up coming along behind my curation.  

RE the discussion above about how incentivize the behavior you want from your curators:
 
I think you can find a middle ground between outright banning curators who only curate the same circle (jerk), and not taking into account diversity of curation in the reward at all.  Something common sense might be generate a curator score from (as first ideas): diversity of authors curated (higher score = less likely to be repeat authors curated over and over); avg. author payout of curated authors (higher score for curating lower rewarded authors); curation reward generated for tipu from curation (which could still be a metric that circle-jerky type curators / curators who curate higher paid authors do well at); curator CSI (giving higher scores for people who are less circle jerky in their overall voting behavior).  Then order curators by their combined score (weighted however makes sense) and scale the payout accordingly.  Circle jerky curators who curate good content (but from a small / well rewarded circle) would fall toward the back third, so they could keep on participating but just not receive as much reward as those who are actually out there actively curating for new content and spreading their vote around.

I am sure you are already thinking of ways to do quality control checks on curators, one easy of course would be to order curators by total negative rshares (flags) received by posts they curate and starting your quality control checks there.