A reader asks, "How long does it take you to "throw" everything in the morning from the previous day's org file?"
How long it takes to process a walklog is completely variable. The main factors are the length of the walklog and the complexity of its headings.
I tend to refile complex headings from other authors without decomposing them, unless I urgently must engage with the subject matter. However, I almost always decompose my own complex headings, unless I can publish them to Pubmind without voltage drop. Even then, I often decompose those published headings to refactor the pieces I still need to scatter to Textmind paths other than the heading's destination path.
I do a lot of my actual work during the decomposition stage of the daylog decomposition checklist. I file the complex headings to ~2-Linked/1-Persinter~ and work on them there, one at a time, using treefactor-org-refactor-heading. It's the last step before I process tasks, since knowing my critical path often requires resolving problems raised in these complex headings.
The mindless mechanical administrative steps of daylog decomposition take a trivial amount of time. But reviewing yesterday's thoughts often sparks further thoughts. I follow these threads until they're resolved, or until the complexity of followup exceeds that which will fit comfortably into the middle of the daylog decomposition checklist. Closing open loops during daylog decomposition saves me the admin overhead of task creation, management, selection, resumption, and review. Plus, it's productive and intelligent to do next-day basic followup.
If rushed, one can defer processing complex headings and just file them in ~2-Linked/1-Persinter~ for whenever. But it's better to close the loop, to ensure one's direction is correct. Leaving complex headings unprocessed risks missing important pivots. Their value is likely to decay, e.g. because one already did it the hard way.
Generally, the longer I spend on daylog decomposition, the more productive the day. The only reason to interrupt the checklist is if I'm mining a rich vein and hitting easy paydirt. If I can get away with it, I'll leave a queue of complex headings in ~2-Linked/1-Persinter~ and decompose them over multiple days. Obviously this isn't safe if they contain urgent info, which is why I put urgent issues into separate headings during heading atomization.
It would be great if I were smart enough to use nothing but the daylog decomposition loop to run my life. I only go beyond the loop when I know I can't trust my brain to handle something. For many people, pure daylog decomposition would be an improvement over their current workflow. It certainly beats sticky notes and Microsoft Word!