Four fixed boxes. Compare each task to the boxes — never to another task.
Is it important? Is it urgent? Answer once, then move on — don't re-read the list.
Important + urgent0
Do next
Important, not urgent0
Schedule at peak gain
Urgent, not important0
Batch or delegate
Neither0
Drop
0 tasks
Why four boxes and not a ranked list. Ranking N tasks means comparing them against each other — on the order of N² comparisons, each one holding two tasks and two criteria in working memory at once. Twelve tasks is sixty-six comparisons against about four available slots, which is why a full list starts to feel uniformly loud. Binning against fixed thresholds costs N comparisons instead, because each task is judged once, in isolation, against boundaries that are already drawn. A coarse ordering you actually finish beats a fine one you don't.
When something changes, requantize — don't re-sort. Take the one task whose situation changed, compare it to the four boxes, move it. Drag it, or delete and re-add. Everything else stays where it is. That's one comparison instead of sixty-six.