**Status.** A working proposal after focused review and response; the player experience remains unvalidated. It works one limited, local discovery across three visits: what the player can observe, what they may reasonably think, what would settle it, and what changes afterwards. Example fiction is labelled. Nothing here answers revival, the station's history or the escape, and no alphabet or language system is designed. These strings may be a fragment of an alien language; this local chain fulfils no whole-language requirement. Relay/sink/isolator service-family notation remains distinct.

*(Proposed fiction.)* **Wall strings**: short runs of cut alien signs, three to five characters, on plain wall surfaces. They belong to the proposed original fabric of the station, distinct from crew paint. The player sees them on *walls* rather than on components; their authorship remains a question. What they say is not proposed here. What the player can learn is where they appear and where they do not.

**Records.** Capturing the facsimile is a world-time action; reading back what is already stored, and inspecting the frozen visible view, happen paused. In the current proposal the journal holds the framed facsimile and its location, while which of the four readings the player favours lives in their own notes, in their own words, and is never promoted to fact by the game. What else notes may carry, and whether records can later be annotated, is [Clues and Help](/Hold%20The%20Flood/Clues%20and%20Help)'s to state; nothing here forbids it.

**Physical states, precisely.** At instance three the panel is *off* and the duct behind is open and continuous. At instance two the panel is *on*, and what lies behind it is unknown until it comes off. At instance one the panel beside the door is on.

**The test the player can now run.** The duct and passage readings both predict something behind instance two. Opening it tests that shared prediction. An unmarked duct can test whether all ducts are marked, but cannot by itself distinguish what a marked wall indicates. This is where the chain stops being observation and becomes a prediction the player can be wrong about.

**The observation that discriminates.** A string beside an already-open removable wall slab, giving onto a short passage with ordinary floor, ordinary wall finish and standing clearance. Directly visible: the string cut in the usual way, the slab, and a passage a person walks along rather than a duct they crawl into.

**Why this, and not an unmarked duct.** The player may also find a duct with no string, welded shut a short way in. That is a real observation and worth recording, but it settles less than it appears: it shows only that marking is not exhaustive, and leaves "a string means a duct behind" untouched, since that reading claims nothing about unmarked places. The stringed walkway is different — the visibly ordinary passage weakens the duct-specific reading and supports the broader practical use of finding a way through.

**What survives.** Matching strings identify candidate wall access. This is a useful practical prediction from the observed cases, not a translation or a guarantee of passability.

**Still ambiguous, with a live rival.** The strings might instead mark what one maintenance team was responsible for, which would correlate with passability without meaning it. Further cases can change confidence: a marked intact surface with no passage and independent maintenance evidence would favour the responsibility reading; marked passages associated with different maintainers would weaken that particular team interpretation. Later damage must be distinguished from a surface never used as access. Those further cases have not been observed in this chain.

**And it reinterprets what was already seen.** The mark beside the door now gives the player a reason to investigate the adjacent panel. Previously recorded strings become candidate access points; their literal meaning remains unsettled.

**What this sample does not establish.** A player may conclude the builders moved through their walls routinely. That is a reasonable reading, but a handful of marked passages is consistent with it and with other explanations, so it stands as the player's inference rather than a fact the chain demonstrates. Nor does the way the work is cut settle chronology.

**After.** The evidence for trying is a sightline, not a hunch. The jammed door has an inspection port, and through it the player can see a distinctive far-side landmark — an emergency lamp beside a painted crew section name. A string sits on a wall short of the door; with that panel off, the same landmark is visible again through a grille deeper in the crawlway.

**The decision.** Take the crawlway, or walk to the door. Nothing is swapped in the loadout to do it: the cutter is part of the suit and cannot be left behind, so no carried tool trades against a module. The choice weighs potential savings in route length, exposure and door-work time against the panel work and the cost of an unsuccessful detour.

**If it fails.** The run pinches out, or is collapsed, or ends at a weld, and the *cause is visible* where it stops, so the player learns something rather than being told no. They have spent the detour's oxygen. Trying the door, retreating or seeking another option depends on current route state, threats and remaining supply.

Failure lowers confidence in the specific route prediction. Visible later damage may explain a blocked passage; without such evidence, other interpretations remain live. A mark alone does not establish that a usable passage once existed.

The chain's inference does not depend on the order in which its evidence is reached. A player who begins at the open panel, where the duct is plain, may first consider the "duct" reading — but it still takes the stringed walkway to move past it, so the same practical inference follows once both are observed. A player who meets the walkway first may consider the broader reading early and compare later crawlways against it. Strings recur at several sites; walking past one does not itself consume the clue. Revisiting a recorded instance requires finding its marked room or surface in the current arrangement. Earlier recorded routes become historical after death. Given the same reached evidence, order changes which hypothesis is held first; routes and access after each shift remain subject to the current arrangement.

Shown even one facsimile aboard, Dorn says the crew called these crawl marks and used them to get around when the main runs were blocked. They never knew what the signs said; they knew what they meant in practice. It is firsthand, true and limited — it does not say what the signs mean literally, which passages are still open, or where this player should go.

**This can happen early, and that is a real path, not a lesser one.** Show him a record after visit one and the practical reading arrives straight away; the panels and the stringed walkway then confirm what he said. A player who never asks derives it from the instances instead. The paths differ in where the reading comes from and in what the later visits do — confirmation against derivation — and both leave the player judging which crawlway is worth the oxygen and what to do about the jammed door. The design does not assume every player would get there unaided, and does not make Dorn cryptic to protect a deduction. He answers what he knows.

Its case is real. It gives the colony a visible role in discovery, and authored dialogue can carry nuance that physical evidence cannot. Three variants are worth separating: help available early and optionally, as proposed here; no help at all, which pushes all the weight onto the instances; and help required, where the reading is unavailable until a particular person is aboard. Only the third creates a gate, and that is a property of that variant rather than of survivor interpretation.

The preference is to preserve an independent route to the practical reading. Required help instead makes that interpretation depend on rescue and dialogue. Either policy can retain subsequent physical confirmation, failed predictions and further reinterpretation; useful testimony can also change expectations.

**Provisional choice:** an available route through physical comparison, with truthful, limited survivor interpretation offered from the first record onward.

**Accepted costs.** Redundant authored instances — several stringed sites, the discriminating walkway, one unstringed welded duct, and at least one passage no longer usable with a visible cause — plus dialogue across more than one knowledge state. Redundancy reduces the chance of a player getting stuck without a way forward; it does not guarantee none, and dialogue does not guarantee it either. None of this is measured.

**What would reopen this.** Evidence that the stringed walkway does not read in play as different in kind from a marked duct. Then the survivor's reading should carry the decisive step, and the honest response is to say so rather than keep a deduction that is not landing.