The answer is not fewer ideas or more typing. It is flow.
The post is right to separate local code speed from system output. Delivery is a chain. Increasing spare capacity at one link cannot raise departures while another link controls the pace.
constraint capacity -> throughput; starts - departures -> WIP; WIP / throughput -> cycle time
Worked example: five times faster Code
After twelve seeded weeks, baseline and 5x Code both ship 2.75 items/week. Code utilization falls below 1%, while 15 proposals wait at Idea validation. Faster typing creates downstream slack, not more validated bets.
Counterexample: elevate Ideas
Two idea validators plus a WIP limit ship 4.75/week and cut cycle time to 1.04 weeks. The bottleneck moves to Review and QA. A real constraint change moves the whole system.
Why 9-9-6 can look busy and ship less efficiently
In the illustrative 72-hour scenario, people start work faster, creative validation saturates, fatigue reduces effective capacity, and review defects loop back to Code. The seeded run reaches 54 WIP, 6.24 weeks of cycle time, and 30% trailing rework while shipping 2.00/week. The exact fatigue parameters are assumptions; the queueing consequence of starts outrunning departures is structural.
Scope: this lab explains a delivery-system mechanism. It does not claim to know why a specific employer uses long hours, measure a live industry rate, or reduce labor conditions to one model.