The Discipline of Variation The Seven QC Tools and Deming’s System View
本話は、物語の流れを一度止め、品質管理の基礎について整理する講義回です。
QC七つ道具とデミングの思想は、単なる技術論ではありません。
それは「問題をどう捉えるか」という姿勢の問題でもあります。
物語を読むための補足ではなく、独立した学びとしても読んでいただければと思います。
Deep Dive
The Seven QC Tools and Deming’s System View
This section is presented as a technical lecture accompanying the narrative.
Quality tools do not create truth.
They restrict what can be claimed without evidence.
When Sakurako opens a QC book, she is not choosing aggression.
She is choosing discipline.
⸻
Part I — The Seven QC Tools
1) Cause-and-Effect Diagram (Fishbone Diagram)
Function: organize hypotheses before collecting proof.
A fishbone diagram maps potential causes (e.g., 4M: Man, Machine, Material, Method) against a defined effect.
It is not evidence. It is a structured suspicion list.
Discipline: define the effect precisely (which line, which dimension, which period).
Common error: turning the diagram into a blame chart.
⸻
2) Pareto Chart
Function: prioritize impact.
It ranks categories by frequency or cost. The 80/20 pattern is common but not guaranteed.
Discipline: standardize counting rules and update after countermeasures.
Common error: hiding major causes inside “miscellaneous.”
⸻
3) Check Sheet
Function: standardize observation.
Data must be designed before it is collected.
Include time, shift, machine ID, lot ID. Without identifiers, later analysis becomes guesswork.
Common error: recording outcomes without context.
⸻
4) Histogram
Function: visualize distribution.
Is the data centered? Wide? Skewed? Bimodal?
A bell-shaped curve does not equal a healthy process.
Specification limits and control limits are different concepts.
Common error: ignoring measurement system error.
⸻
5) Scatter Diagram
Function: explore relationships between variables.
Correlation suggests inquiry, not causation.
Non-linear effects and thresholds matter.
Common error: mistaking association for proof.
⸻
6) Control Chart
Function: distinguish common-cause variation from special-cause signals.
It is a decision rule, not decoration.
Control limits are derived from stable data and indicate statistical stability—not customer acceptance.
Common error: “tampering” with the process after every fluctuation.
⸻
7) Stratification
Function: reveal hidden structure.
Data may appear stable only because different conditions are mixed together.
Separate by shift, machine, supplier lot, or setup condition.
Common error: attempting stratification without identifiers recorded in advance.
⸻
Part II — Dr. W. Edwards Deming
Dr. W. Edwards Deming (1900–1993) influenced modern quality thinking, particularly in postwar industrial practice. His contribution was less about formulas and more about philosophy.
1) “In God we trust. All others must bring data.”
This maxim is not anti-trust.
It operationalizes trust.
Vague managerial language—“commitment,” “attitude,” “effort”—cannot replace measurable definitions.
⸻
2) “94% of problems belong to the system.”
The percentage is symbolic.
The principle is structural.
Most defects arise from system design:
•process capability,
•equipment limitations,
•material consistency,
•policy decisions,
•training structures,
•incentive systems.
Default hypothesis: examine the system before blaming individuals.
⸻
3) Implication for Process Capability (Cp, Cpk)
Process capability indices describe variation relative to specification limits.
If Cpk declines, the disciplined question is:
•What changed in the system?
•Was the process stable?
•Did measurement conditions shift?
•Was stratification performed?
The question is not immediately moral.
It is structural.
⸻
Concluding Perspective
The Seven QC Tools do not accuse.
Deming’s philosophy does not comfort.
Together, they impose constraint.
They limit the range of acceptable explanations.
For Sakurako, the book is not a weapon.
It is a framework (枠組み) that converts unease into testable structure.
And structure, once defined, demands evidence.
ばらつきは、悪ではありません。
重要なのは、それをどう理解し、どう扱うかです。
感情や印象で説明することは簡単ですが、構造として検証するには規律が必要です。
今回の講義が、物語の背景理解だけでなく、実務や学習の参考になれば幸いです。
ご感想があれば、ぜひお聞かせください。
