--- slug: mystery-analysis type: pattern summary: "A sufficiency-logic method for learning from the gap between an expected outcome and what actually happened by finding and correcting the assumption behind the prediction." created: 2026-11-01 updated: 2026-11-01 related: assumption: relation: uses note: "Mystery Analysis searches beneath a failed prediction for the assumption that made the original reason appear sufficient." sufficient-cause: relation: uses note: "The method expresses a hypothesis as an if-then claim and tests whether its premise, action, and reason were sufficient for the predicted outcome." categories-legitimate-reservations: relation: uses note: "The Categories of Legitimate Reservations test the original and replacement explanations against the observed result and their predicted effects." learning-surprising-events: relation: contrasts-with note: "Both methods learn from an expectation gap; Mystery Analysis is a compact logical analysis, while Learning from Surprising Events is an eight-step team inquiry that proceeds through policy change." continuous-improvement-auditing: relation: complements note: "An S&T audit can surface a mystery when reality contradicts a predicted effect; Mystery Analysis supplies a focused way to examine the failed hypothesis." --- # Mystery Analysis > **Pattern:** A named solution to a recurring problem. *Structure a failed prediction as a hypothesis, use sufficiency logic to find the assumption that no longer holds, and replace it with an explanation that accounts for the observed outcome.* **Reasoning mode:** [sufficiency-based logic](sufficient-cause.md) — arrows read *if A, then B*; multi-in-edge defaults to OR. The name describes a specific learning move, not an unsolved puzzle in general. Eli Goldratt used *mystery* for an outcome that exposes the limits of current reasoning. Yuji Kishira later named a Thinking Processes method for examining such outcomes **Mystery Analysis**. The unexpected result supplies the evidence; the mystery is the assumption that made a reasonable prediction fail. ## Understand This First - [Assumption](assumption.md) — the belief or condition that explains why a cause-and-effect relationship is expected to hold. - [Sufficient Cause](sufficient-cause.md) — the if-then logic used to state and test the hypothesis. - [Categories of Legitimate Reservations](categories-legitimate-reservations.md) — the scrutiny used to challenge both the old explanation and its replacement. ## Context Mystery Analysis begins with a prediction that was explicit enough to be contradicted. A practitioner believed that an action, under stated conditions and for a stated reason, would produce an outcome. Reality produced something materially different. The difference may be a failure, but it can also be an unexpectedly good result. Either one shows that the working explanation was incomplete. Three related uses of *mystery* need to remain separate. Goldratt used the word broadly for an unexpected outcome that creates an opportunity to learn. In Strategy and Tactics Tree auditing, a mystery is an observed effect that contradicts a predicted effect of the tree. Kishira's Mystery Analysis is the named process for structuring and examining the hypothesis behind such a gap. The method is Kishira's; the broader learning idea and the word are not. The method is narrower than [Learning from Surprising Events](learning-surprising-events.md). Both start with an expectation gap and look for a flawed assumption. Mystery Analysis gives one practitioner or a group a compact logical structure for one prediction. Learning from Surprising Events adds event selection, a mixed inquiry team, explicit handling of blame, competing explanations, policy revision, and organizational sharing. ## Problem When an action produces an unexpected result, hindsight makes a weak explanation feel obvious. People may decide that the action was wrong, the person executing it failed, or the result was random. A favorable surprise is even easier to wave away as luck. None of those responses tests the reasoning that produced the original expectation. The prediction usually contains more than the visible action. It also contains a premise about the situation, a reason the action is expected to work, and one or more assumptions connecting that reason to the outcome. If those elements remain implicit, the practitioner can't tell which part reality contradicted and may repeat the same reasoning under a different label. ## Forces - **The original prediction may have been reasonable.** One contrary result doesn't prove that every part of the reasoning was false. - **A plausible explanation can fit the result without causing it.** The analysis must look for predicted effects that distinguish competing explanations. - **Several causes may produce the same outcome.** Stopping at the first candidate turns hindsight into certainty. - **Better-than-expected outcomes also matter.** Treating success as confirmation can hide the assumption that would make the success reproducible. - **The evidence may remain incomplete.** A corrected hypothesis can be better supported without being proved. ## Solution State the prediction as a four-part hypothesis: **premise, action, reason, outcome**. The premise describes the relevant condition. The action says what was done. The reason explains why the premise and action were expected to be enough. The outcome states the predicted effect in observable terms. Then place the actual outcome beside the prediction and inspect the gap. 1. **Write the prediction as it existed before the result.** Avoid repairing it with facts learned afterward. 2. **Check the premise and the action.** Establish whether the stated condition held and whether the action actually occurred. 3. **Surface the assumptions inside the reason.** Ask why the premise and action were expected to produce the outcome. The answer is the load-bearing explanation. 4. **Generate alternative reasons.** Include explanations for both the missing predicted effect and the effect that appeared instead. 5. **Apply the Categories of Legitimate Reservations.** Test causality existence, cause insufficiency, additional cause, and predicted effects. Ask what else would be observable if each candidate explanation were true. 6. **Replace the reason only as far as the evidence permits.** Prefer the explanation that accounts for the observed result and its predicted traces. Record unresolved alternatives rather than converting survival into proof. 7. **State the revised prediction.** The new reason should predict the result that occurred and make a future observation possible. This procedure is not Kishira's unpublished seven-question presentation reconstructed from its title. His 2016 TOCICO talk used that description, but the question list isn't available in the accessible sources. The procedure here follows the hypothesis structure and analysis described in his 2018 article and the later TOCICO Body of Knowledge note. The Mystery Analysis sufficiency diagram below shows the default form for candidate explanations. Each candidate alone might be enough to produce the observed effect, so the convergence uses no junctor: ```mermaid flowchart BT H1[Demand changed after the prediction] --> O[Actual lead time exceeded the prediction] H2[A supplier delivery arrived late] --> O H3[Priority rules interrupted the planned work] --> O ``` The default diagram keeps independent explanations visible. The override form is needed when the claim depends on several entities acting together rather than on any one of them alone. The second Mystery Analysis sufficiency diagram shows the hypothesis structure as joint sufficient cause. The premise, action, and reason must all hold together before the predicted outcome follows, so the diagram uses an explicit AND-junctor: ```mermaid flowchart BT P[Premise: materials are available] --> J(("AND")) A[Action: release the order early] --> J R[Reason: earlier release gives the order more time] --> J J --> E[Predicted outcome: the order finishes earlier] classDef andjunctor fill:#A6894B,stroke:#3a2f1a,color:#fff,stroke-width:1.5px; class J andjunctor; ``` ## How It Plays Out **An early release makes delivery later.** A planner releases an order early because all materials are available and expects the extra time to improve delivery. The order instead finishes late. Saying "the shop was busy" doesn't yet explain the result. The analysis states the original premise, action, reason, and expected outcome, then tests what early release changed. If the release enlarged the queue, increased priority changes, and delayed work at the constraint, those predicted traces support a revised reason: in this environment, releasing earlier can add work-in-process and interference without creating more constraint capacity. The revised prediction can then be tested on the next comparable order. **A change works better than predicted.** A team alters one operating rule and sees a larger improvement than its hypothesis allowed. Declaring victory loses the lesson. The gap may reveal that the rule affected a second causal path, removed an unrecognized conflict, or changed behavior at the constraint. Mystery Analysis treats the favorable result as a challenge to the old reason, then asks which additional effects should appear if each candidate explanation is correct. These are illustrations of the method, not measured outcome claims. The available Kishira sources explain the logical structure and argue that it speeds learning discussions, but they don't report comparative performance data for the method. ## Consequences **Benefits.** Mystery Analysis turns a surprise into a test of reasoning instead of a search for fault. Separating premise, action, reason, and outcome prevents an implementation error from being confused with a faulty causal claim. The CLR supplies a shared vocabulary for challenging explanations, while the revised prediction makes the learning testable rather than merely persuasive. Applying the same discipline to favorable surprises can expose conditions worth preserving. **Liabilities.** A single event rarely identifies one mechanism by itself. Candidate explanations may share the same predicted effects, evidence may be missing, and several causes may have acted together. The method can therefore produce anti-learning if the practitioner treats the last surviving explanation as proved. It also depends on a prediction specific enough to contradict. If the original outcome was stated as "things will improve," there is too little structure to analyze. Mystery Analysis doesn't replace a [Current Reality Tree](current-reality-tree.md) for chronic, connected undesirable effects or a team process for a consequential organizational event. It fits the bounded case: one expectation, one observed gap, and a hypothesis whose assumptions can be surfaced and tested. ## Sources - Yuji Kishira and Roy Stratton, ["Mystery Analysis"](https://www.tocico.org/_files/ugd/1ed33e_5d550a6095c643c08e71060d5d751d56.pdf), TOCICO-recognized Body of Knowledge note (2021). Source for TOCICO's definition, the distinction between Goldratt's general use of *mystery* and Kishira's named process, and the identification of the Thinking Processes and CLR as its machinery. - Yuji Kishira, ["An Examination of Goldratt's Theory of Constraints as a Scientific Theory: Using the Thinking Processes to Propose the 'Structure of a Hypothesis' and the Mystery Analysis Processes"](https://journalcra.com/sites/default/files/issue-pdf/29374.pdf), *International Journal of Current Research* 10(4), 68443–68450 (2018). Primary treatment of the premise-action-reason-outcome structure, the application of Mystery Analysis to a prediction gap, and the worked Kaizen example. The article is used as an author-originated account; no claim about peer review is made here. - Goldratt's Golden Nugget 13, ["Reaching the Plateau of the U-Curve"](https://www.toc-goldratt.com/tocWeeklyNews/2011/tw37.php), *TOC Weekly* (page undated; it promotes a January 2011 event). Source for the Strategy and Tactics audit use of *mystery* as an observed effect that contradicts a predicted effect. - Eli Schragenheim, ["Systematic Learning from Significant Surprising Events: Drawing the Right Lessons for Improving the Future"](https://www.tocico.org/_files/ugd/1ed33e_8e54ccede7304bdab6903589df2504b0.pdf), TOCICO-recognized Body of Knowledge paper (2022). Source for the neighboring eight-step team method and the boundary drawn between that process and Kishira's compact hypothesis analysis. --- - [Next: Strategy, Marketing, and Sales](strategy-marketing-sales.md) - [Previous: Assumption](assumption.md)