Tool 25 of 32 · an afternoon
Gate-First Sequencing
Anything whose outcome could change other work is a gate, and gates run first no matter how small they look.
In Plan, on the candidate task list, after the depth verdict and the ranked moves exist.
What it is for
To order the work so that anything whose outcome could change other work runs first, regardless of size, and so that subtraction leaves the backlog in writing.
The failure it catches
Ten days of building, then the two-day measurement that changes the requirements, then ten days of rework. Measurements never look urgent because their value is what they teach, not their size.
Why it works
One question, if this task's outcome surprised us would it change how we do other tasks, finds the work whose value is informational. Scheduling that work first, however small it looks, is what prevents the expensive loop of building, then measuring, then rebuilding.
How to run it
- For every candidate task, ask: if this task's outcome surprised us, would it change how we do other tasks.
- Anything that qualifies is a gate. Gates get scheduled first, regardless of how small they look.
- Anything a pending gate could invalidate waits behind it, or is parked in writing with its price.
- Anything that should not happen this quarter goes on a parked list, in writing, with the price of parking it. In leadership's head nothing ever truly leaves the backlog, so it leaves in writing instead.
- Write weather assumptions for any plan that depends on a sunny day.
- Tag every plan item: gate, waiting behind a named gate, or free to run.
How to read the result
Every item in the plan carries one tag. It is a gate, it is waiting behind a named gate, or it is free to run. An untagged item is an unexamined risk. Everything marked gate is the same as nothing marked; a gate must name what it could invalidate.
How to run it well
- A gate names what it could invalidate. Size is not the test.
- Parking is a decision with a price. The price goes on the sheet next to the item.
Where it breaks
- Everything marked gate is the same as nothing marked.
- Foundation-physics claims can shelter slowness. The claimed setting-phase must name its condition, or it is just a longer estimate.
- Skipping here because planning feels like progress, before the pattern is confirmed, produces a guess with a schedule.
What would prove it wrong
A gated plan that still produces the ten-then-measure-then-rework loop means gate identification failed.
What we watch
Whether gated plans stop the build-measure-rebuild loop, or whether a gate we failed to mark keeps showing up as the two-day measurement that invalidates ten days of work.
Ask yourself
- If this task's outcome surprised you, would it change how you do other tasks?
- Which items in the current plan are waiting on a measurement you have not run yet?
- What is on the parked list, in writing, with a price?
- Does every plan item carry one tag: gate, waiting behind a named gate, or free?
Fed by: Deadline Replay, Contradiction Gate, Trade-Off Price Tag