Bracken Vale · Essai 02

Which sentences can stay?

The steering paper recommends a temporary ramp. Its cited drawings and prices are unchanged. Then a cheaper bridge becomes available. Can the old recommendation keep its reviewed status?

The steering-pack assistant keeps the wording and source snapshot for a whole draft. This separate worked case gives the board six sentences to judge. Two already lacked sufficient evidence before the new option arrived. Compare a whole-paper review, a rule checking only the old citations, and a fresh calculation of each claim’s declared meaning.

The previous paper and its current grounds
Sentence in the previous paperWhole paperOld citationsMeaning checked againReason

In the two baseline columns, “review triggered” means the method calls for another review; it does not mean the sentence is false. In “Meaning checked again”, “Retain” means the same sentence is still justified for the same use after the typed evidence is recalculated. “Reopen” means the supplied facts now defeat the sentence. “Needs review” means conflict, a scope overclaim or an undefined judgement prevents retention.

Read the old and current evidence together

·

This reader selects fixed examples. The link names the current snapshot and reload restores it. Its controls create no project record. The old paper, each changed packet and every calculated disposition remain available here. Authoring or reviewing a real steering paper belongs to a separate workflow.

The unexpected source of staleness

A sentence recommending the least-cost choice relies on the set that was searched. Its evidence includes the absence of a better admissible option. A new option can defeat the sentence while every old citation stays current. Try “An unusable option arrives”. Lift X costs only 1 unit and needs an attendant who has not been authorised. The ramp can still be the cheapest admissible option. The board has to check the lift’s conditions to reach that verdict.

These are counts of sentences referred for human attention under eight authored changes. Every typed rule is recalculated. Reading the sources, resolving versions and entering the facts are work supplied to the model. The count therefore measures neither total effort saved nor a general ability to review prose.

Which fact should the board commission next, and when can the review stop? Follow the worked assurance decision in What must we reread?

At a different station, change the evidence at Fellbridge to see which options remain eligible. Then select “Open the full rehearsal”. Choose an approach and give your reason. After the site walkthrough, change or reaffirm your choice and read your two reasons side by side.

What another reader supplied

A second agent independently specified the six-claim Alderbrook challenge and four changes before seeing this implementation. We used that challenge as the basis of this Bracken Vale model. Its 24 expected dispositions agree with the calculation. This is a design challenge supplied independently, not a second case or blind outcome test.

A later adversarial review introduced four night shifts where installation estimates meant uninterrupted days. The scalar comparison was inadequate: productive hours, restart overhead and permission to split the work were unknown. The revised model marks both installation feasibility and option choice unresolved until a common time basis is supplied. It does not invent a conversion.

Read the independently supplied challenge · Open the reproduced result · Follow the reasoning, work comparison and sources