What Should Be Built
The observatory exists to describe systems as they are. But the question underneath it is not only descriptive:
Which works would actually relieve the constraints that bind the systems carrying global flows?
Lists of desirable infrastructure already circulate, and their weakness is always the same: they enumerate ambitions rather than relieve identified constraints. This register answers the question under conditions that make the answer checkable — or refuses to answer it.
An intervention must be earned by a bottleneck
Corridor record → bottleneck register → intervention
An intervention that cannot name the documented bottleneck it targets is not a weak intervention; it is not an intervention at all. The constraint has to be established independently, before anyone thinks about the remedy. That single rule is what separates this register from a wish list.
Permanently a scenario
The claimed effect carries epistemic_status: scenario, fixed by the schema. Building the work does not convert the claim — that requires new evidence on the corridor metric. Construction shows something was built, not that the system improved.
Displacement is required
Every entry names the constraint that becomes binding once the targeted one is relieved. Widening a segment usually moves the bottleneck; a proposal that does not say where it moves is incomplete, not optimal.
No priority score
Priority is argued in words against documented criteria and selected by a human. A number would hide the criterion the argument actually turned on.
The register
Two tiers, and the difference is visible on every entry. A candidate is a structured hypothesis about where a constraint lies and what would relieve it — the analytical to-do list, honest about being one. A validated intervention requires a validated corridor record, at least one evidence reference and sufficient evidence. The schema enforces both conditions on the status field, so the tiers cannot be blurred by an author in a hurry.
| Intervention | Targets | Next binding constraint | Type | Tier | Evidence |
|---|---|---|---|---|---|
| Publish a shared ferry slot schedule across the Caspian crossing and align berth allocation at both shores, so that a booked crossing has a known departure window rather than an open wait. | bottleneck:caspian-crossing-coordination | port handling and rail path availability at the western interface | coordination_change | candidate — not citable | 0 · insufficient |
How priority is argued
A priority_rationale is expected to engage with:
- how binding the targeted bottleneck is, and for which flow families
- how many segments or functional links the relief would reach
- whether the displacement is smaller than the relief
- the strength and independence of the supporting evidence
- feasibility preconditions — financing, jurisdiction, permitting, operator agreement
- whether a cheaper coordination or regulatory change would achieve the same relief
What this does not authorise
- Ranking works across corridors whose records are not comparable in scope.
- Publishing an intervention whose only evidence is its sponsor’s own material.
- Aggregating interventions into a global investment figure.
- Presenting a candidate without its non-citable notice.
The full reasoning is in Dossier 6, and the field contract in the register specification. Entries are authored in data/public/interventions/ and validated against the schema on every push.