I know subjectivity is hard to code but if there was some way to include higher rewards based on not just keywords but relevance that could help eliminate spam. Maybe there could be some way to provide negative incentives for using #boolberry related keywords without any relevant context
Like a Proof of Constructive Human Work?

I still like that idea the most, and "to be eaten food" is kinda easy to proof to be constructive human work. Living in a house shaped as the BBR logo would be even better proof

...
Rather than using coding to determine relevance (and higher rewards), allowing users to give/ take away tips by upvoting (or downvoting) relevant (or irrelevant) tweets would be more in keeping with the game. The premise of Popboolr is to reward users based on what other users deem 'popular'.
There may be a need for an algorithm to filter spam, if spam becomes a major issue. However that bridge can be crossed once it's reached.
That being said if someone is able to code a 'Proof Of Constructive Work' algorithm, that will enhance the user experience, then I'm open to explore it! Having a daily pot where users can upvote/ downvote where it gets distributed certainly sounds doable.
CZ, could bigj have the hard fork but in a way that BBR has e.g. 1% of mined limited to 1 year for a purpose? Miners should agree on this, it will distribute BBR to humans through a game which adds value at this age of BBR, and 1 year limits any inefficiencies that may/will develop for the process?
In my view, it is undesirable (and unnecessary) to build rewards for promoting BBR into the protocol itself, even for a limited period of time.
Incentivising devs to continually improve the code base I understand. However an online game like Popboolr can (and should) have a model that is self-financing or (worse case) donation based rather than having its funding built into the currency.