Clarifying insight... from better experts.
Are you tired of seeing wonderful insights in research talks, and knowing they won't reach introductory content for years? I suggest the importance of direct research expertise, in creating accessible and powerful introductory content, is pervasively underestimated.
K-graduate... because the hard part is upstream.
An MIT professor writing a children's picture book puzzles over whether world phytoplankton population is best characterized as predation or resource limited. Even with grad student support, it's a hard question. Once answered, writing assorted versions for K and graduate is the easy part.
Rough-quantitative... active learning?
Rough-quantitative reasoning might provide a medium for collaborative inquiry and non-toy analysis, and thus for misunderstanding pruning and play. This would require uncrippling estimation best practice, K-graduate. And creating rough-quantitative descriptive content.
Content improvements... are very not independent.
I'm suggesting aggressively combining existing threads of content improvement effort, and asking "Look where they're headed, years from now. Years of glacial incremental creep. Isn't it wonderfully pretty there? Isn't it worth paying a high transient cost, to get there this decade rather than next?"
Tasty low-hanging fruit... abundant & neglected.
Poor incentives, crippling constraints, and daunting deployment challenges... science education improvement is hard. But these also cause opportunities to be overlooked. Surprisingly large opportunities. This site looks at accessibly teaching size and deep time. More generally, having a vision of where we want to end up, helps turn up tasty near-term opportunities.