7 September 2026 addition: the temporary-power loop is a new fourth essay, using the Python MCDP package to size a cyclic component system. It retains all three survivors and the retirement decisions below. Model and source.

6 September2026. Project Co-design · Categorical-method audit

The active collection contains staged paths, programme studio, and wildlife crossing. The studio is repaired; wildlife is rebuilt. Three weaker rail apps are retired as standalone calculations. Their useful ideas remain in the surviving essays rather than depending on an executable predecessor.

Preservation map

Earlier explanation Surviving home What is preserved or corrected
Incremental upgrade: a package provides functionality while requiring resources; induced signalling/power requirements must agree. Studio coupling primer and its package/compatibility explanation. The connected-parts account is retained. Eligibility now uses one consistent requirement; no18tph package is endorsed for a36tph request.
Rail simulator: several resource-incomparable choices can remain, and stakeholders need a discussion of trade-offs rather than a forced single score. Studio coupling primer, its resource frontier and director’s reading. Retains the governance discussion role. Modelled constraints are not represented as actual stakeholder consent; a frontier slope does not prove its external cause.
Transit/incremental: one commitment can force further subsystem work. Studio coupling primer and staged method, “Requirements may change. Commitments are another choice.” Preserves the practical lesson while distinguishing an order-monotone design relation from irreversible committed assets. Exploratory requirements may be lowered.
Wildlife: a crossing needs guiding infrastructure and monitoring, not just a bridge. Rebuilt wildlife essay. Retains the scenario and interdependence, now with actual interface joins, mixed bridge bundles, all declared resource costs and a frontier of complete witnesses.

The studio’s prescribed-strategy comparison remains separate from the staged solver’s constructed paths. The wildlife model remains an ecological-design toy; it does not inherit railway assumptions or claim observed animal response.

Why the older rail calculations are retired

  • Rail simulator: selected packages could fail its own signalling/power requirements; every default highlighted resource choice failed its power condition. The studio covers the worthwhile package/programme comparison.
  • Incremental upgrade: catalogue choices could contradict the displayed throughput/power settings. Its advertised antichain was not computed. The coupling explanation was more valuable than maintaining another solver-like front door.
  • Transit trade-offs: thin duplication plus the misleading claim that mathematical monotonicity means never lowering an exploratory slider.

Their original addresses now explain the withdrawal and lead to the preserved explanation or surviving model. Historical source is retained at the pre-curation revision. Old saved settings are not silently passed off as equivalent inputs to another model. The five generic-gallery addresses still reach their matching dedicated routes, and no retired cards are reintroduced to those catalogues.

Repairs to the survivors

The programme studio removes undisclosed180-month/260-risk rejection thresholds, evaluates its specified affine feedback relation, distinguishes catalogue infeasibility from user ceilings and provides an explicit baseline implementation where the current railway already meets the task. Its base construction pattern is distinguished from programme-level delay.

Wildlife replaces a fixed recipe staircase with a finite component catalogue and compatibility composition. It includes mixed bridge widths, costs fences and monitoring—including staff—and retains the incomparable external resource vectors with their full witnesses. The old two50m recommendation at120 model crossings/day is beaten by a30m+50m bundle with the same support packages. That counterexample is now a regression case.

The staged solver’s verified mathematical engine is preserved unchanged. Its source and method explanations now link to the surviving essays, and its monotonicity/commitment explanation incorporates the useful lesson from the retired sliders.

Why this review went beyond the migration checks

The earlier migration correctly established that files were published, links and controls worked, and several narrow defects were repaired. That did not prove every catalogue choice was feasible, every displayed bundle minimal, or every empty result justified. The critical review found counterexamples. This curation preserves the sound ideas, removes the failed calculations and adds model-level checks alongside navigation and mobile checks.