Collection
Change the system, not the slogan
Generate and test structural changes by exposing causes, contradictions, constraints, buffers and adjustable parameters.
Explore in any order.
What’s included
2 checklists · 3 heuristics · 3 protocols
8 ideas
Map possible causes before picking one
A cause diagram is useful because it delays the jump from symptom to favourite explanation. List plausible cause families, attach observations, and mark what is still only a hypothesis before you decide what to test.
Read the idea ChecklistSimplify before you automate or delegate
Automation can make a bad process fail faster. Delegation can move confusion to someone else. First remove work that does not need to exist, then standardize the parts that truly repeat, and only then decide what should be automated or handed off.
Read the idea ProtocolBenchmark the mechanism, not the surface
Copying a successful system is easy; knowing which part caused the advantage is harder. Compare constraints, mechanism and evidence before importing the visible practice.
Read the idea HeuristicReverse the default to find an option
When every idea is a variation of the current approach, reverse one assumption on purpose. The reversed idea does not have to be good; its job is to expose what the default was hiding.
Read the idea HeuristicUse the ideal outcome to reveal costs
An impossible ideal can still be useful if you treat it as a lens. Describe the result you would want if the harmful cost disappeared, then work backward to identify which cost, dependency or function is truly necessary.
Read the idea ChecklistReuse a proven resource before adding one
New work often begins by recreating a template, component, dataset or explanation that already exists nearby. Search for reusable parts first, but carry their assumptions and rights with them.
Read the idea HeuristicPut buffer where variation hurts
A buffer is useful when uncertainty would otherwise propagate into failure. Place time, capacity or inventory around the unstable boundary, then measure whether the buffer absorbs real variation or merely hides a broken process.
Read the idea ProtocolChange one parameter, then observe
When you change several parts of a routine at once, success feels good but teaches little. For reversible low-risk changes, vary one meaningful parameter, define the signal you care about, and compare before deciding what to keep.
Read the idea