If you are closer to the codebase and deep code handling (math pole), you are farther from its actual application (art pole) and vice versa. This happens because you have limited time which allows you to be good in some specific area. Getting full stack is costly in development, getting wide range is costly in production.
I tried to depict such a dependency here.
Switching between tasks take efforts, getting deep into a task (obtaining a full context) take efforts.
In my opinion, prioritizing the tasks (task ranking) is also a separate management job that connects devs and industry requirements, so you have to be familiar with both of them at quite deep level, to distribute the amount of overall care properly.
But managing wide range require mastering wide range first.
In this position, you will have too much power and the ability to screw up a lot of things if you are not qualified in some areas (for example, when an animator familiar with maya is trying to design modeling workflows, confusing system design with features design or usability with low entry threshold), so this power is distributed among the modules.
Slower but safer?