Appendix: evidence scope and a field mechanism challenge

The evidence architecture distinguishes institutional guidance, fictional model inputs, reproduced outputs and original analysis. Guidance supports a narrow proposition about a method or process within its original scope. Fictional inputs support an illustration. Reproduced outputs support the mathematical consequence of those inputs. Original analysis connects mechanisms, challenges and decision frameworks. None should be promoted into another evidence class simply because the paper is attractively rendered or mechanically long.

A source card identifies the public publisher, retained edition or accessed page, locator, supported proposition and limits. Retrieval evidence establishes that readable primary material was obtained and retained. It does not automatically establish that every sentence in the manuscript is supported. The author therefore keeps source-derived notes narrow and separates them from the original system argument. Independent review should inspect the precise locator and the consequence drawn from it, not merely count bibliography entries.

The evidence cutoff defines what this edition had available. A live page can change after retrieval, while an archived PDF preserves a particular document object. A later source revision should trigger a support review if it affects a consequential claim. The paper should retain the earlier receipt and explain the change rather than silently replace the evidence beneath unchanged text. Edition transparency concerns actual source and model changes, not an invented history of reviews that never occurred.

The study defines no empirical operating universe. It therefore asserts no observed bottleneck distribution, cargo failure frequency, regional climate probability or market tariff. Its public institutional sources supply methodological context, not measured parameters for the fictional chain. This exclusion protects the reader from an easy but unsupported transition: using a credible publisher's broad statement to make a precise local capacity claim that the publisher never established.

The historical World Bank logistics report and its redesigned indicator page have different evidence bases. A perception-based assessment summarises a defined respondent population and methodology. Operational shipment indicators summarise another kind of observation. A change between their displayed results cannot be interpreted as an unchanged time-series movement without a bridge. The paper imports neither score nor ranking; it uses the contrast to explain the need for evidence-object and edition definitions.

A practitioner should ask what the external measure actually observes, which period and population it covers, how missingness is treated and what decision it supports. A national context indicator can be useful for framing diligence, while a corridor movement record can test a local mechanism. They can inform the same discussion without becoming interchangeable. The bridge should state the relationship and limit of each source rather than combine their authority into an unsupported numerical conclusion.

The same principle applies to maritime outlook evidence. The retained UN Trade and Development source excerpt supports qualitative uncertainty and disruption mechanisms only. Direct full-report retrieval was unavailable in the underlying receipt, and no forecast quantity or market statistic is asserted here. This limitation remains visible. A later successful full-report retrieval would permit additional locator review, but would not retroactively validate a claim that this edition did not make.

Use a broad indicator or outlook to identify the local service question, then collect operating evidence capable of quantifying that service. Declare those inputs before the annual app computes their consequences. An external report cannot validate them through citation, and respected institutional context is not an empirical baseline for an unnamed corridor. Keeping the evidence bridge explicit lets the reader trace which source framed the question, which local record supports the input and which conditional result the calculation supplies.

An expansion sponsor can point to a congested loader, a larger design rating and the importance of port access without demonstrating that loading governs annual delivered flow. The controlled cases below retain the declared serial service and expose the distinction. They ask which capacity change can alter the governing minimum when every other stage remains on its stated operating basis. That question is narrower than whether the port intervention is operationally valuable, financially attractive or necessary for a particular vessel window. Its answer gives the team a mechanism to investigate before committing to a causal explanation.

The loading-only case retains the original annual flow and rail constraint. The rail-only case moves the binding stage to vessel loading and gains little annual flow because loading then governs the minimum. The coordinated case reaches the stockyard constraint. These contrasts do more than name the initial bottleneck: they expose how an apparently large local expansion can produce a small system change and identify the next constraint that a follow-on proposal must address. Because the operating costs were deliberately held fixed, the lower unit cost in a higher-flow case is an allocation consequence. It is not a verified efficiency saving or return on investment.

A measured shortfall at rail would still require a causal investigation. Intrinsic capacity can be low because of traction, train length, track occupancy or infrastructure condition. Observed flow can also be low because the mine releases irregularly, documents are incomplete, the stockyard cannot receive the train or loading operations make acceptance unavailable. Those are different interventions. A rail utilisation ratio based only on annual effective capacity cannot distinguish them. The field record must classify productive work, intrinsic loss, starvation and blocking with consistent timestamps and reasons, preserving uncertain classifications for review instead of treating all idle time as rail failure.

The observation contract should define the movement object before teams compare records. Identify a consignment or train across mine release, dispatch, receiving acceptance and onward loading. Record the event definition, local time basis, source system and any correction history. Link documents and quality acceptance to the same movement where the operational question requires them. If one team measures gate departure and another measures accepted receipt, their durations can be combined only after the boundary is reconciled. A missing link should remain a missing link; it should not be replaced with a convenient average that makes the chain appear continuous.

The accepted-service denominator also needs attention. A chain may move physical tonnes that are not yet available for a vessel because quality, documentation or receiving conditions remain unresolved. A port operation may receive cargo earlier and therefore improve service reliability without increasing annual tonnes. The annual model measures its declared physical throughput and costs; it does not silently value documentary readiness or schedule reliability. The committee should state whether it is authorising greater annual flow, more reliable dated service, lower operating cost or a combination, then retain separate evidence for each promised outcome.

A useful controlled operating challenge changes the hypothesised binding mechanism while observing the surrounding interfaces. For example, an agreed dispatch trial could alter the release discipline while keeping the receiving procedure stable. The record would track whether idle rail time shifts from starvation to productive work and whether queues or refusals move downstream. A maintenance intervention would need a different protocol that distinguishes restored intrinsic capability from more favourable arrivals. Neither trial is a causal proof merely because flow increases afterward. The interpretation should account for cargo mix, demand, weather, authority changes and other concurrent influences relevant to the declared service.

The decision response is a staged authority packet. First establish the service and a coherent baseline. Then identify which evidence could distinguish the plausible mechanisms and who can supply it. Use the annual contrasts to bound the structural question and a suitable dated model where windows, variability or inventory govern the intended benefit. A staged trial can precede an irreversible capacity commitment when its result would change the design or scope. Keep the option to stop visible, including the cost and operational burden of further evidence. Repeated analysis without an actionable distinction between outcomes does not justify deferring a decision indefinitely.

A successful intervention should carry an explicit duty to examine the next constraint. Increasing productive rail capability while leaving loading unchanged moves the governing annual stage rather than abolishing the serial minimum. Joint rail and loading capability can then expose stockyard receiving or handling. The operating team should prepare acceptance rules, dispatch interfaces and feasible contingency arrangements for that transition. A later constraint is not necessarily evidence that the earlier intervention failed; it can be the expected consequence of relieving the earlier limit. The sponsor should distinguish this planned progression from an unexpected interface failure that invalidates the assumed service.