# Enough for one workable choice

PW-225 · Essai 02 · 10 October 2026. **Executed finite analysis; practical benefit untested.** Baseline: `24b32e5ee3713517ed75bb19f4867653a4182eed`.

**Making a dependable action available can remove the need to find out which situation you are in.** Here, the service route is known usable in every admitted layout; the remaining layout uncertainty no longer blocks this route choice.

The previous whale essai made a route selectable by adding information. This trial holds that route criterion fixed and adds a known usable service passage. The handler can then choose service without discovering which old passages are open. This does not identify every option, and it says nothing about whether the subsequent physical move finishes.

## What was actually checked

Let A(w) be the supplied goal-admissible actions in world w. A deterministic local rule is constant on each observation group C. It can meet the checkpoint requirement in every admitted world exactly when the intersection of A(w) over the whole C is nonempty. Full option identification instead requires equal sets A(w) throughout C. Selecting a common action gives the elementary sufficiency argument; any successful uniform rule must choose such a common action, giving necessity. Neither criterion is a general theorem about temporal coordination.

| Arrangement | Worlds / partitions | Minimum outcomes for one choice | Minimum outcomes for all options |
|---|---:|---:|---:|
| Original east/west menu | 3 / 5 | 2, with two alternatives | 3 |
| Known robust service route | 3 / 5 | 1 | 3 |
| Service fails in west-only layout | 3 / 5 | 2 | 3 |
| Service setup uncertain | 6 / 203 | 2 | 6 |
| Changed criterion requires original passage | 3 / 5 | 2 | 3 |

The uncertain-setup case admits both setup values for each layout. Enumerating all partitions is a theoretical information comparison, not a claim that every partition has a realizable sensor. The priced intervention menu implements only the stated route cues, full briefing and known service setup. No setup observation is quietly supplied.

Default illustrative effort prices favour a cue costing 1 over service costing 3, although service requires fewer observation outcomes. If service costs 1 and a cue 2, service wins. At equal price both cues and service tie. All 16 menu subsets are checked and every cheapest package retained. Full briefing costs twice a cue. These weights are authored, not measured effort or information entropy.

Known service availability includes applicability and authorisation premises. Unknown setup cannot be wished away by selecting the true ready world: indistinguishable not-ready alternatives remain. The changed-goal case explicitly excludes service despite physical availability; it is a new assessment contract. No intervention here supplies authority.

The original completion model remains separate and unchanged. A potentially missing forward report with acknowledgement still leaves reopening and robust nested knowledge unresolved. Known accurate inspection or known reliable delivery resolves the supplied checkpoint. A selected route is not evidence of completion. We add no retries, termination argument or simulation of movement.

## Source → concept → construction

Primary sources reopened on 10 October 2026, using the linked full HTML/PDF passages:

- [Bolander 2017, arXiv v1](https://arxiv.org/html/1703.02192), Definition 7 and §9 Definitions 9–11: applicability at the actor's perspective, uniform choice, successful finite execution. Our implementation is the finite checkpoint intersection, not DEL product update or a temporal strong-policy planner.
- [Engesser et al. 2017, arXiv v1](https://arxiv.org/html/1703.02196), §3.3 Definitions 9–11: ownership, knowledge of preconditions, local determinism and uniformity. We do not claim arbitrary replanning converges or reproduce their perspective-sensitive search.
- [Moses 2016, arXiv v1](https://arxiv.org/html/1606.07525), §2 Definition 2.1, §3 Theorem 3.1/proof and asynchronous footnote: local-state knowledge and necessary known preconditions for the stated conscious actions. Necessary knowledge does not grant success or authority; our simple proof is separate.
- [Halpern and Moses 1990](https://groups.csail.mit.edu/tds/papers/Halpern/JACM90.pdf), §§5–6, PDF pp10–13: local histories and view-based knowledge. The retained completion check uses observation alternatives, including missing delivery. No common-knowledge claim is made.
- [Hutchins 1995, author manuscript](https://pages.ucsd.edu/~c8johnson/COGS102B/Hutchins95.pdf), manuscript pp18–20, cockpit-system discussion: material and social arrangements can change cognitive work. This motivates the service alternative; the invented passage is our construction, not an empirical conclusion from the cockpit study.

## Independent checks and reproduction

`model/options.test.mjs` independently supplies boolean action tables, generates every partition by canonicalising all label functions, and enumerates every deterministic local rule. It agrees with intersection synthesis on all 223 scenario partitions. A three-action counterexample additionally rejects the tempting but wrong pairwise-overlap criterion. Existing ten tests retain all 128 original arrangement/channel comparisons and nested-knowledge controls. New six tests cover the comparison, prices, failures and completion boundary.

```sh
node --test model/*.test.mjs
node model/export-options.mjs --check
```

[results.json](results.json) contains every partition, common action set, example rule and obstruction, not only the summary. The page renders the same module. Regenerate with `node model/export-options.mjs`; no network or dependencies are needed for these checks.

## Running evaluation

The useful result is a distinction between two demands that can sound alike: preserving the option set and retaining enough to act. The added action reduces the latter information requirement without reducing the former. It also separates information complexity from effort cost. This is a checked instance of an elementary distinction, not a novelty claim.

The ordinary action table captures the entire substantive result. The mathematics earns a modest place by enumerating all alternatives and checking the observation restriction, not by finding an otherwise inaccessible project insight. The service route's availability, knowledge, cost and authority are supplied; this trial has not reduced dependence on the modeller designing the possibilities. It advances the End by making a physical-arrangement repair comparable with information work while exposing that limit. No practitioner decision, human comprehension or realised project benefit has been tested.