GOLD STANDARD · THERMODYNAMICS · HOW X WORKS
Find the decision hidden inside the model
A marked near miss can end in recovery, maintenance, shutdown or failure, and voluntary logs observe only part of that pathway. A gold-standard exergoeconomic shield tests underreporting sensitivity, models competing transitions and compares interventions through explicit counterfactuals while keeping safety constraints outside the price calculation.
This guide connects the Gold Standard of Thermodynamics, Gold Standard of Education, and How X Works hub.
Chapter contents and reading routes
Choose the route closest to your present task. Every chapter remains expanded, so the article also reads naturally from start to finish.
Orient the event process
Model competing transitions
Build intervention counterfactuals
Transfer and govern
Use evidence and next routes
CHAPTER 1 OF 12 · ORIENT THE PROCESS
1. A near miss begins a pathway, not a single count
After a shared cooling anomaly, the system may recover, enter maintenance, shut down safely or progress toward damage. These outcomes compete because observing one changes or prevents the others.
Preserve time, operating state, mechanism and severity marks. A total report count discards the transition evidence needed to decide whether an intervention changes recovery, merely changes reporting or redirects risk into another outcome.
CHAPTER 2 OF 12 · ORIENT THE PROCESS
2. Separate event occurrence from event reporting
A voluntary report appears only when an event occurs, is noticed, classified and submitted. Reporting can depend on mark severity, crew, workload and whether an intervention already attracted attention.
Use alarms, work orders, audits or mandatory records to estimate reporting sensitivity where possible. When it is not identifiable, declare a range and propagate it through the decision rather than assigning one precise correction.
CHAPTER 3 OF 12 · MODEL TRANSITIONS
3. Model competing transitions on a state diagram
Define states such as normal, stressed, degraded, maintenance, recovered and failed. From stressed, recovery, maintenance and failure are competing exits. From maintenance, restoration and extended outage may compete.
Estimate transition hazards or probabilities with exposure time and censoring visible. A cause-specific hazard is not the same as the cumulative probability of an outcome, because the other exits remove opportunities for that transition.
CHAPTER 4 OF 12 · MODEL TRANSITIONS
4. Work through an underreporting sensitivity range
Suppose 18 stressed-state near misses are reported: 12 recoveries, 4 maintenance transitions and 2 shutdowns. Audit evidence suggests reporting sensitivity of 0.60 to 0.90 for recovery marks and 0.90 to 1.00 for the other marks.
Expanded recovery events range from 13.3 to 20, while maintenance ranges from 4 to 4.4 and shutdowns from 2 to 2.2. The latent mark distribution is therefore less severe than the raw report distribution, but total event exposure remains uncertain.
| Reported transition | Reports | Sensitivity range | Expanded range |
|---|---|---|---|
| Recovery | 12 | 0.60–0.90 | 13.3–20.0 |
| Maintenance | 4 | 0.90–1.00 | 4.0–4.4 |
| Shutdown | 2 | 0.90–1.00 | 2.0–2.2 |
CHAPTER 5 OF 12 · MODEL TRANSITIONS
5. Validate transitions against an independent stream
Compare predicted state occupancy and exit timing with sensors, inspections and maintenance timestamps that were not used to construct the report marks. Check calibration of each transition and dwell-time distribution.
If stressed-state reports predict maintenance but not sensor deterioration, the model may be learning reporting practice. Independent evidence helps distinguish an operational transition from an administrative one.
CHAPTER 6 OF 12 · COMPARE INTERVENTIONS
6. Define the intervention counterfactual precisely
Ask what would happen under a declared alternative such as earlier inspection, reduced load or immediate fallback, at the same starting history. The counterfactual must specify how the intervention changes transitions and operating costs over time.
Observed outcomes after intervention cannot simply be compared with untreated reports when assignment depends on severity. Use a justified design or causal model, including measured time-varying confounders and the possibility that intervention changes competing exits.
CHAPTER 7 OF 12 · COMPARE INTERVENTIONS
7. Compare risk and exergoeconomic consequence together
Suppose ordinary operation has estimated 30-day probabilities of 0.70 recovery, 0.20 maintenance and 0.10 shutdown. A load-reduction intervention changes them to 0.78, 0.18 and 0.04 but adds 40 units of exergy-cost loss.
The intervention reduces shutdown risk by 0.06. Compare economic consequences only among options meeting the hard safety requirement under the full underreporting range. If the upper shutdown bound remains unsafe, a favourable average cost cannot pass the shield.
| Strategy | Recovery | Maintenance | Shutdown | Added exergy-cost |
|---|---|---|---|---|
| Ordinary | 0.70 | 0.20 | 0.10 | 0 |
| Load reduction | 0.78 | 0.18 | 0.04 | 40 |
| Difference | +0.08 | -0.02 | -0.06 | +40 |
CHAPTER 8 OF 12 · TRANSFER AND GOVERN
8. Changed-context transfer: data-centre cooling
Move from an industrial heat system to redundant data-centre cooling. The stable structure is selective incident reporting, marked transitions, competing exits and a safety-constrained intervention. Workload, service-level losses and component architecture change.
Transfer succeeds when states, exposure time, reporting evidence and intervention dynamics are rebuilt. A shutdown in one setting may be a protective success; in another it may be the primary service failure.
CHAPTER 9 OF 12 · TRANSFER AND GOVERN
9. Use sensitivity surfaces, not one corrected number
Vary reporting sensitivities, transition models and intervention effects together. Plot whether each candidate remains inside the safety envelope and how economic ranking changes.
Focus new evidence where the decision boundary is steep. If all credible combinations favour the same safe action, more parameter precision has little operational value.
CHAPTER 10 OF 12 · TRANSFER AND GOVERN
10. Limits and common misconceptions
Near misses are mechanism evidence, not failures. They should not be added mechanically to failure counts, nor discarded because harm was avoided. Competing-risk models organise transitions but do not identify causal intervention effects by themselves.
Underreporting adjustment can increase event counts without changing exposure information. A rate still needs the relevant operating time or demand denominator and a stable event definition.
CHAPTER 11 OF 12 · CHECK AND CONTINUE
11. Questions reliability teams often ask
Use a multistate model when intermediate states and repeated transitions matter; use a simpler competing-risk model when one initial state leads to mutually exclusive first outcomes. Match complexity to the decision and validation data.
Can voluntary reports estimate absolute risk?
Only with additional assumptions or independent evidence about reporting and exposure. They are often stronger for mechanism discovery and relative pattern checks.
When is a counterfactual credible enough for a shield?
When the intervention is well specified, assignment and time-varying confounding are addressed, competing outcomes are included, and safety conclusions survive reporting and model sensitivity.
CHAPTER 12 OF 12 · CHECK AND CONTINUE
12. Evidence, specialist routes and the next check
NASA ASRS guidance on voluntary-report limitations explains why raw reports cannot establish prevalence. research connecting multistate models and causal interventions clarifies the assumptions needed for transition counterfactuals, while counterfactual prediction under competing risks shows that interventions can change both the target outcome and its competing events.
Continue through near-miss reporting bias and state validation, time-varying common-cause updating, and policy shielding and safe fallback. The next check is a single sensitivity surface joining reporting, competing transitions, safety and exergoeconomic cost.
Contents · Previous chapter · Return to the Thermodynamics root