Maintenance is hard.
Writing new code, developing a new algorithm, establishing a new routine. That stuff is fun. The novelty drives a certain amount of progress. But eventually the shine wears off and maintenance kicks in.
You’ve lost the weight, now keep it off. You’ve published the paper, but now folks are emailing you with clarifying questions. This type of work pulls you away from the next new thing, but it is critical. Maintaining a foundation ensures that the future things you build and attempt have sure footing. If your next new thing is to rebuilt that which you let crumble, then you’re spinning your wheels.