i disagree with the first point. I'm working on a distill article with Chris and Shan, and the major draw for this has been impact. It seems very plausible that an article on distill has the potential to reach a far broader (and different) audience than a paper in even a top tier mathematical journal like SIAM would.

I won't deny the time commitment needed for a distill article is not trivial - it is far more work than a technical blog. But in terms of a pure tradeoff of time per publication, the calculus makes sense. Most of the work of research distillation and synthesis is already part of the research process, and writing a distill article is just a matter of putting it all of down on paper. Doing research is a far more time consuming and less predictable process.