A "problem" is the gap between expectation and current state. Suppose the expected value of something is B, and the current state is B′, then the gap (B′→B) is the problem. ... To solve this problem, you shouldn't focus on B′→B; instead, you need to look through B′ at ABC, which I call the "Dialysis Prism". ... Step 1: Calibrate Goal B: The goal must comply with the SMART principle, and avoid treating means as goals. Step 2: Reconstruct Method A: To change the current state, don't add a new solution based on the current state; instead, go back and reconstruct the original method system. Step 3: Eliminate Variable C: Use the basic framework of "Xiang (Phenomenon), Shu (Data), Li (Principle)" to find it, and dig out the real cause through the 5Why questioning method. —— Chapter 17: Dialysis Prism

1. Definition of problem: Problem = Expectation(B) - Current State(B'). Without identifying the gap, you can't find the problem; "no problem" often means not knowing the goal or the current state
2. Surface problems are symptoms B'→B, while root causes lie in three factors: A (Method), B (Goal), C (Variable)
3. Prism B (Calibrate Goal): First check if the goal itself is correct. Common mistakes: vague goals (e.g., "happiness" is unmeasurable), treating means as goals (e.g., "earn 1 million" may be a means rather than a goal)
4. Prism A (Reconstruct Method): Current state B' is caused by the original Method A. Treating the symptom means finding a new path from B' to B; addressing the root cause means reconstructing the original method system itself
5. Prism C (Eliminate Variable): If both A and B are correct but B' still exists, there must be a hidden Variable C. Use the "Xiang-Shu-Li" framework to find it: Phenomenon → Data → Principle. Use 5Why consecutive questioning to dig out the root cause
6. Key principle: Don't give advice too early — "We love giving advice too much, but often haven't figured out what the problem really is"
7. Essential difference in methodologies: The Self-Strengthening Movement was patching the old system (treating symptoms), while the Meiji Restoration was reconstructing the entire system (addressing root causes)

• Self-Strengthening Movement vs Meiji Restoration: The Qing government only learned Western technology without changing the feudal system (patching the old Method A), and eventually failed; Japan reformed both the system and technology (reconstructing the entire Method A), and became a great power overnight — the huge gap between treating symptoms and addressing root causes for the same problem
• Factory production decline during the World Cup: The phenomenon is production decline (Xiang), the data is that the night shift defect rate soared 3 times (Shu), and 5Why questioning revealed: Workers stayed up late watching the World Cup → poor mental state → operational errors → increased defects (Li). Improvement actions focused on "things we can change": Prohibit bringing mobile phones into the factory
• Analysis of rising complaint rate: Instead of subjective statements like "I feel complaints have increased", use data — "2‰ last month vs 6.7‰ this month, a 3-fold increase", shifting from perceptual cognition to rational analysis
• Mistake of taking "happiness" as a goal: Happiness is unmeasurable → can't find the gap → don't even know what the problem is → can't start solving it

• The same problem recurs, emerging again shortly after each "solution"
• Multiple solutions have been tried but failed, and you start to doubt if the problem definition itself is wrong
• Need to conduct root cause analysis (review, accident investigation, performance improvement)
• When formulating improvement plans, want to ensure you grasp the essence rather than the surface
• Team discussions fall into a cycle of "treating symptoms only"

• "Why does this problem happen again?"
• "Why can't we solve it for good?"
• "I've tried so many methods but none work"
• "What exactly is the root cause of this problem?"
• "Where exactly is the problem?"
• "Are we always treating symptoms instead of the root cause?"

• vs. brain-unseal: Brain unseal is a preliminary step to restore brain operation capability, while Dialysis Prism is a problem analysis tool after the brain is functioning properly. Sequence: Unseal first → then use the prism
• vs. structured-thinking: Structured thinking is a tool for disassembly and classification, while Dialysis Prism is a tool for penetrating surface to find root causes. You can first use structured thinking to fully break down elements, then use the prism to check each one
• vs. mirror-world: The prism applies to deterministic problems (ABC can be identified), while mirror-world applies to uncertain problems (ABC itself is uncontrollable)

• Write down the problem you are currently facing, using the format "B - B' = Problem": What is the expectation (B)? What is the current state (B')? Where is the gap?
• Check if Goal B complies with the SMART principle: Specific (S), Measurable (M), Achievable (A), Relevant (R), Time-bound (T)
• Check if you are treating "means" as "goal": Ask yourself "What will I get after achieving B?" If the answer is the real goal, replace B with it
• Completion criteria: Goal B can be expressed with a specific number or clear verifiable state, and it is the final objective rather than an intermediate means

• Sort out which links (people, processes, tools, rules) are included in the original Method A that leads to the current state B'
• Instead of adding a new solution based on B', go back to A itself and ask "Which link in the original method caused B' to deviate from B?"
• Conduct systematic reconstruction rather than incremental patching for that link: Replace instead of叠加
• Completion criteria: Can draw a comparison diagram of "Method A before reconstruction" and "Method A' after reconstruction", and A' is not a patch added to A, but a change to a core link

• If the problem still exists after reconstructing A, start variable investigation:
  • Xiang (Phenomenon): What specific abnormal performances have been observed? List all abnormal phenomena
  • Shu (Data): Find quantitative data to replace "I feel" — What is the data? How much is the change range?
  • Li (Principle): What is the reason behind the data change? Use 5Why to ask at least 5 consecutive times until finding the root cause you can change
• Focus improvement actions on "things you can change" rather than uncontrollable external factors
• Completion criteria: Produce a data-driven problem analysis report (Phenomenon → Data → 5Why chain → Root cause → Executable improvement actions), and 100% of the improvement actions are within your control

• Entirely new fields with no historical Method A to trace back to — No old method to reconstruct, need to design from scratch
• Purely uncertain events (stock market crash, sudden disaster) — The ABC model is not applicable, switch to the four-domain strategy of mirror-world
• Emergency accidents requiring immediate action — No time for in-depth analysis, stop the bleeding first then review
• The problem itself is simple enough that the cause is obvious at a glance — Over-analysis wastes time

• Skip B and directly look for A: If the goal itself is wrong, all subsequent analysis is based on a wrong foundation
• Patch A instead of reconstructing: Add a new rule/process every time a problem occurs, the system becomes more bloated but the problem still exists
• Stop 5Why at the first level: e.g., "Production decline → worker error → end", without asking why the worker made the error
• Find the root cause but can't control it: e.g., "Because the World Cup is being held" — The analysis is correct but no action can be taken, which is useless
• Replace data with "I feel": Without quantitative analysis, the prism degenerates into random guessing

• Lacks priority guidance on "how to determine whether to start troubleshooting from B, A or C", in practice all three directions are possible, requiring a lot of time and resources to check one by one
• The 5Why method relies on the experience of the questioner, different people may derive completely different causal chains from the same phenomenon — No calibration mechanism
• "Reconstructing Method A" may involve complex factors such as organizational structure adjustment and interest redistribution in practice, which is far more difficult than described in the book
• The Xiang-Shu-Li framework comes from the Book of Changes but is simplified to "Phenomenon → Data → Principle", in actual analysis, determining "Li" often requires professional domain knowledge support

• structured-thinking (prerequisite): The analysis process of Dialysis Prism (calibrate B, reconstruct A, eliminate C) requires structured thinking as a basic tool. First use the MECE principle to fully break down problem elements, then use the prism to check each root cause.