Let's say you have a queue processor called Q. It takes in jobs described by tuples of (hours, utility) like (8,1), (2,4), (5,15).
Using the above example, you make an interesting observation: if your queue processor processes only two items, it has an efficiency of (1+4)/(8+2) = 0.5 utilities per hour. If you process three items, it shoots up to (1+4+15)/(8+2+5) = 1.33 utilities per hour.
Remarkable! It would seem that the more items on the queue you process, the more efficient this queue processor becomes. Perhaps you have just discovered the Law of Increasing Marginal Queue Processing Utility.
Or perhaps the queue is being processed in the wrong order.
Or perhaps we attach a third field to the tuple and call it a "priority" and require that everything with higher priority is done before anything with low priority. Let's call it (I am probably inventing the term here) "priority queue".
Author actually writes about it. There are plenty of tasks that may be of low value to him but are absolutely essential for external reasons.
High-value items are often dependent on low-value* things
* low value on their own. Breathing/sleeping/eating/commuting/booting-up is a very low value activity, but it is a prereq for all high-value activities
Using the above example, you make an interesting observation: if your queue processor processes only two items, it has an efficiency of (1+4)/(8+2) = 0.5 utilities per hour. If you process three items, it shoots up to (1+4+15)/(8+2+5) = 1.33 utilities per hour.
Remarkable! It would seem that the more items on the queue you process, the more efficient this queue processor becomes. Perhaps you have just discovered the Law of Increasing Marginal Queue Processing Utility.
Or perhaps the queue is being processed in the wrong order.