Hold The Horde — Story And Ending Map

This page proposes one causal spine, four mystery and personal chains, and player choices across apparent, deeper and optional-secret layers; the “leading branch” is the team's concrete version for the owner to inspect and revise. Owner decisions are recorded in the Owner Brief.

Three levels of authority

The causal working spine

1. Ancient intervention. The builders create a field that can restore damaged living form. In the leading captive branch, the experiment leaves a living ascendant organism dependent on or confined by the field. Builders isolate the organism from the wider repair template and abandon the station for an unknown reason. Their ordered runs, access markings and containment machinery survive; none explains motive.
2. Human hospital. Humans encounter extraordinary healing and build ward machinery, records and transfer systems around it. Researchers open or bypass the old isolation to increase reach and edit treatment profiles. Some active hospital machines begin repairing tissue toward a mixed, changing profile. Patients, staff decisions and hostile creatures are consequences of those particular interventions, not proof that every revival corrupts.
3. Present return and shifting. The astronaut's berth records one active return relationship. Restoration returns the body and worn/carried property to that registered ship-side point; the same intake transaction causes unbound station fabric to re-seat while persistent joins/work remain. This supplies a cause for the selected gameplay rules without changing them.
4. Player intervention. Regional evidence lets the astronaut separate ordinary restoration from harmful treatment profiles, trace one disputed human transfer and reach containment. The player may prepare to leave, isolate the mixed profile, leave it alone, accept/refuse a voluntary treatment, or keep investigating. No free-text interpretation or language completion secretly selects an outcome.

Ordinary corruption rule recommended for this branch. Use evidence-only ordinary revival plus the voluntary treatment bargain. Death itself adds no mark, budget, forced transformation or ending state. A treatment can offer a visible ancient-interface benefit and a visible hospital-machine hazard before consent; refusal preserves every main path, and one known ship purge removes both benefit and hazard without upkeep.

Compact competing branch — the empty core/template engine

The strongest alternative keeps the ancient experiment and hospital intervention but removes the living captive: the ascendant left, while a machine retained its form template. Humans treated that template as a patient and altered it, producing the harmful mixed profile. The physical content changes: the deeper cell needs an inspectable template lattice and stored profile comparisons instead of living containment evidence, and responsive utterances become reproduced procedure rather than possible address to a prisoner. The creative cost is material: the strongest image—someone deliberately sealed inside—disappears, along with the possibility that one physical act also answers a confined party.

Its deeper choice also changes. Instead of re-seat the isolation around a living source or leave it coupled, the player chooses which already evidenced repair profile the engine should use—or leaves it untouched. A cleaned human baseline offers a colder engineering decision; an ancient template risks repeating an unknown experiment. No person, including the astronaut, is permanently bound or transformed by this alternative. A sacrificial personal template would be a different proposal and is excluded here.

Four mystery and personal chains

Chain 1 — the berth that remembers

First discrepancy. After an ordinary death, the astronaut wakes at the ship with still-carried objects in their exact surviving states, while a Fork deliberately installed at Cargo Turn remains there and completed fasteners remain released. The station arrangement changed; time did not rewind. The berth's intake display carries the same recurring subject token and current body-response trace seen at the first arrival.

Rivals. The ship made a copy; the field relocated one body; time reset only the astronaut; a general station repair system happened to choose the berth. Aster's belief that arrivals disturb the station is reasonable testimony, not proof.

Discriminating observation/action. On a normal expedition the player may leave the Fork installed at objective 1 and record one completed fastener, then later compare those facts after an ordinary death that occurs in play. No deliberate death is required. The same object remains installed, the fastener stays open, consumed oxygen stays spent and the carried manifest returns once. At a ward record, a historical patient restoration occurred at that patient's bed rather than at the astronaut's berth. A later direct berth diagnostic can show that the current return trace is registered to this docking arrival, but cannot explain identity or consciousness.

Correction and limit. Persistent world state rules out a whole-world rewind. Different restoration locations weaken “every repaired body goes to this ship.” They do not distinguish relocation from reconstruction, prove the field's range or establish a survivor rule. A route record from before death is historical even when its object/work facts persist.

Later use, cost and human consequence. The player can predict that reclaiming a station-installed part is a different commitment from dying with a carried one, and can evaluate an irreversible departure honestly. Aster may ask to compare her arrival theory with the berth trace. Sharing it lets her retract “every crossing causes a shift” while retaining “this docking may”; refusing or postponing costs no relationship or help. The chain costs one optional diagnostic capture and a return comparison.

Chain 2 — the harmful treatment profile

First discrepancy. Early ward logs show repaired injuries without the later correction pattern. A later human treatment record introduces a profile identifier also visible on one active ward machine; a creature from that ward carries a matching tissue pattern. Dorn remembers treatment after the seals opened, but his changed perception is evidence about him, not a causal verdict.

Rivals. All resurrection eventually corrupts; the core field changed by itself; humans installed a harmful profile; the creature's pattern is unrelated disease; the record and creature merely share a ward label.

Discriminating observation/action. In a clinical treatment fixture, the player places two equivalent living culture patches from the same recorded source behind a transparent shield—one under the machine's active profile and one with the named profile physically bypassed. This is a voluntary test on non-sentient tissue that runs while the game runs. The active side restores toward the visible anomalous pattern while the bypassed side closes without it. Removing power or interrupting before both committed observations yields an incomplete test; it does not prove safety. Objective 6's line trace and objective 9's regulator state can then identify which hospital branch carries that profile toward the core coupling.

Correction and limit. The comparison supports “this human-configured profile causes this observed pattern under these conditions” over “all revival necessarily does.” It does not prove the origin of every creature, that the core is innocent, that changed people are less themselves, or that field restoration is metaphysically safe.

Later use, cost and human consequence. The player can bypass the profile at a ward, re-seat the deeper isolation if the leading branch is selected, or knowingly accept/refuse the optional resonant treatment after both its access response and hostile correction cue are visible. Bypass work costs time, exposure and possibly a reusable part installed locally; voluntary treatment changes route/loadout reactions and remains purgeable without upkeep. Yeva wants the harmful line isolated. Dorn asks the player not to call every change damage and supplies his own account. Mara can preserve one completed culture trace or use a separate expendable sample in her material project; preserving evidence and consuming salvage are distinct choices. None withdraws help if the player chooses differently.

Chain 3 — the last transfer

First discrepancy. The R-3 wall pressure record marks its last stable reading with a mechanically impressed pressure-loop cycle value. Yeva dates that history to before the final evacuations; Aster's supply ledger records a canister and restraint shipment later; Dorn recalls nobody crossing after he sealed the receiving section. All three can be honest and incompatible, and their testimony exists only if that person is aboard and shown the record.

Rivals. One clock drifted; Aster recorded a planned rather than completed shipment; the line was redirected; another person used the same identifier; Dorn is hiding a transfer; shifting geography has been mistaken for chronology.

Discriminating observation/action. At objective 6 the player restores the sampled pressure line, follows physically visible source/destination evidence or the long route, and opens a record drawer at that destination fixture. Its transfer card carries the completion impression from the same pressure-loop cycle counter whose earlier value is physically impressed on the R-3 wall record. The exposed ratchet advances when the line cycles and has no observed reverse control; the destination value is later in that mechanical sequence. Visitor-pair knowledge can predict the destination, but gauges and the long approach remain sufficient. A failed continuation, unpowered line or missing current access records an incomplete attempt, not a false witness.

Correction and limit. The shared counter supports one narrow order: this apparatus completed a transfer after the R-3 record's last stable pressure cycle. It does not reconcile Aster's and Yeva's calendar clocks; clock drift remains a rival for their dates. Counter replacement, reset or a copied false impression also remain possible until separately evidenced. The result establishes neither who travelled, why Dorn sealed the section, whether the patient survived nor which account is morally right. A shift changes current access, not the captured sequence values.

Later use, cost and human consequence. The physical chain needs only captures of the wall impression and destination card/counter; no survivor is required. The destination becomes one investigation/rescue lead and its line state helps choose a pressure-control route. Back aboard, the player decides whom to show. Only a present, participating person speaks: Yeva may withdraw a specific accusation that Dorn falsified the shutdown without forgiving the seal; Aster may admit her ledger cannot identify a passenger; Dorn may reveal that he sealed the section after hearing its cycle, not seeing who crossed. An absent or unselected person contributes nothing. Postponement causes no unseen conversation or relationship penalty. The chain changes a relationship and next action without solving a culprit.

Chain 4 — what the core contains

First discrepancy. The objective-9 console updates after physical work and acknowledges the correct fitting's B plate even when the fitting is untouched. That establishes a response protocol but is equally compatible with a job system or an interlocutor. Builder seals point inward; human notes call what lies beyond patient, engine and prisoner in different hands.

Rivals. A living captive sends or receives the lines; an automatic console performs addressed procedure; an empty engine reproduces a departed language; humans projected a person onto machinery. The utterance correlation never proves life.

Discriminating observation/action. Objective 11 opens one passage to a deeper evidence cell. Common evidence before branch selection includes the physical seal coupling, a repair-profile feed and a sight baffle the player can open while the game runs. The story branch the owner chooses determines what direct observation exists behind it:

• Leading captive package: the baffle exposes living tissue continuously exchanging material with the containment line. Direct instruments show recurring biological activity independent of console text. This establishes a living organism, not divinity, sapience, consent, authorship or the meaning of an utterance.
• Empty-template package: the baffle exposes an inactive chamber and a lattice replaying stored anatomical patterns with no observed living tissue. This supports an engine/template reading but cannot prove nothing living exists elsewhere.

These packages are mutually exclusive.

Choice, revisit and limit. In the leading package, the player may re-seat the old isolation, leave the coupling open or postpone after inspection. Re-seating is mechanical work done while the game runs: the coupling tell goes dark and the harmful profile feed disconnects. Ordinary ship revival remains active; continued exploration does not become mortal. A later ward revisit is required to observe whether its treatment profile changed—no remote “safe” verdict appears. Leaving keeps current behavior and future choice available. Whether the prior line asked for isolation remains unresolved, and sealing a dangerous system may coincide with an inferred request without proving it.

Human consequence. Ivo frames voluntary treatment as body agency: accepting, refusing and purging are choices he respects, while he objects to hidden exposure. Mara wants one profile trace preserved before isolation because completed evidence, unlike a live feed, survives disconnection. Yeva sees a chance to stop the ward machinery; Dorn refuses the claim that restoring an older shape restores a prior person. Aster favors leaving the seal choice open until the berth/shift evidence is better. These positions change later conversations and what each person asks the player to examine; they do not gate operation, friendship, equipment, language or an ending.

Beat, evidence and action layers

“Apparent,” “deeper” and “optional-secret” are discovery layers, not a number or hierarchy of endings.

Historical evidence and intervention availability

The optional diagnostic is not a history reader. The limited chronology comes from three records sharing explicit cross-references:

1. At objective 11, a trapped seal-service strip sits physically beneath the later human adapter. Its builders' cut style and punched value are visible; the overlay supports “seal service before this adapter,” while ancient authorship remains an inference.
2. Chain 2's human profile-commissioning card carries one field transcribing that same punched value as found, then names the installed human profile revision. A mistaken transcription or replaced strip remains possible.
3. Objective 1's berth intake trace carries the active profile-revision field and the directly observed docking registration event. Matching the human revision supports “profile commissioning before this docking,” not a universal calendar conversion.

Together, physical layering plus the two cross-references support old seal state → human profile installation → current docking registration for these records only. The player compares them in the paused journal; no interface declares the chronology true. A whole-station diagnostic later measures only the current pulse at those discovered endpoints and each present transformation. It can show a live coupling, a bypass or a break; it cannot recover the trapped strip, read an old event or infer intent.

Intervention before construction. Before re-seating isolation, the player may capture the live coupling tell and complete Chain 2's active-versus-bypassed culture comparison. Re-seating commits the disconnected profile state: stored captures persist, while the live mixed-profile signal and coupling tell disappear. Objective 12 can later connect every room physically without restoring that signal. Its optional diagnostic reaches the present disconnected boundary and reports no transfer there. If the player never captured the live feed, no pre-isolation record appears. A culture comparison attempted only afterward makes both equivalent patches close without the anomalous pattern under the remaining ordinary repair input; that is evidence of the current disconnected state, not proof of what the active feed did before.

Construction before intervention. Visiting objective 11 and completing objective 12 does not operate isolation. With the coupling still open, the completed physical network and optional diagnostic may expose and record the current mixed-profile path. Re-seating isolation afterward leaves every mechanical join and physical-completion count intact but breaks that diagnostic branch at the visible seal boundary. Historical records and prior captures remain.

Reopening boundary. Isolation is deliberately reversible from the same reachable approach-side release. Before re-seat, the control previews disconnect mixed profile; ordinary revival unchanged; reversible here. Reopening happens while the game runs, with a visible warning that it reconnects the mixed profile; the coupling tell and live signal return only if their source is currently active. It does not erase completed evidence or automatically harm anyone, and it may allow a missed current test on a later reachable visit. Continued access after a shift is not guaranteed. Leaving isolation closed can therefore preserve a clear evidence gap without a hidden ending lock, mortality change or regenerated clue.

Compatibility with both regional orders

• Build-and-ship first: objectives 1→2→3→4→8→6→10 expose departure readiness and much of Chains 1–3 before the formal language work 5→7→9→11→12. Objective 6's gauges/long route obtain the transfer evidence without language; later runs/pairs reinterpret how the line was redirected. The player may inspect readiness, postpone and pursue the deeper branch.
• Language-and-network first: 1→5→7→3→6→9→4→8→11→2→10→12 makes the pressure/transfer and reply protocols available before objective 11. Rescue can occur whenever its short geometry exists. Objective 10 later exposes the same departure choice; no earlier core action commits it.

In both orders, an unreachable current room defers its chain while another eligible objective remains. Captured records persist across death with historical provenance; present access, changed machine state and live route claims require current observation. No next shift is guaranteed to expose the desired room.

Human scene obligations within five people / fifteen scenes

The chains imply five short scene variants, intended to overlap home follow-through rather than exceed five people and fifteen scenes:

1. Aster compares one berth trace and narrows—not abandons—her arrival theory.
2. Yeva and Dorn see the completed profile comparison and disagree over whether changed form equals harm.
3. Mara asks whether to preserve a completed culture/profile trace before a live feed is disconnected; this is evidence stewardship, not her salvage-project consumption.
4. Yeva, Aster and Dorn confront the late-transfer record; each revises one claim, and no damaged recipient field becomes a solved identity.
5. Ivo states the voluntary-treatment consent boundary before the player accepts/refuses/purges; later response reflects the chosen act without friendship being a reward.

Cast And Short Rescue Map allocates these variants within its fifteen primary scene briefs, and Campaign Evidence And State Crosswalk maps their participant and evidence conditions. If a named person is not selected, the evidence function can move to the stated pressure-tech, supply-clerk, maintenance, reclamation or rehabilitation role without making that role mandatory.

Reuse and a smaller cut

The chains and layers reuse objectives 1, 6, 9, 10, 11 and 12 and add no regions, objectives, modules, rooms, language chains, people, primary scenes or major placed mechanisms. Room hosts for the collar/service intake, clinical ward/pressure destination, regulator gallery, iris/deeper cell and core area are listed in Station Room And Module Inventory, and the reversible core isolation/coupling is listed in Encounter And Mechanism Placement Inventory.

Smaller concrete cut. Remove Chain 3 as a multi-step personal investigation. Keep the late-transfer card as one unresolved objective-6 clue, with no destination drawer and no three-person confrontation. The causal spine, both languages, departure, harmful-profile and core choices survive. What is lost is the main case where evidence changes relationships without solving the central mystery: Yeva cannot retract a specific accusation, Aster's ledger never meets a physical result, Dorn's seal remains generic suspicion, and home becomes more about explaining machinery than living with incomplete human consequences.

Recommendation, costs and owner choices

Strongest case. Use the leading captive branch, evidence-only ordinary revival plus voluntary treatment, and a deeper isolation choice that disconnects the harmful mixed profile while ordinary ship revival continues. It preserves the owner's strongest image, lets one physical act matter both mechanically and ethically, gives every chain a practical consequence, and never requires language, a survivor or a death count to progress.

Accepted creative costs. A practical reason to seal can overshadow the uncertain request; making treatment safer may weaken later horror; cautious players can keep corruption external; direct living-tissue evidence still cannot establish personhood; and five knowledge-state scene variants plus mutually exclusive core content require careful placement. The empty-template branch remains a coherent colder interpretation with a more technical final choice, while the responder may still be only a console even in the captive branch.

Two owner choices that genuinely require taste:

1. Core image: choose the leading living captive behind isolation (recommended), or the empty template engine. The first preserves a party the player may answer but risks anthropomorphising an unknown organism; the second makes the intervention a profile-selection problem but loses the prisoner image.
2. Apparent departure: choose voluntary undocking as an explicitly irreversible apparent-credit action that ends field revival and station exploration (recommended), or keep readiness as a repair milestone with no departure/credits yet. In the recommended version, postponement remains indefinite and no player becomes mortal while continuing to explore.

Open questions

• Core image: the living captive behind isolation or the empty template engine (owner choice 1 above).
• Apparent departure: voluntary irreversible undocking, or readiness as a repair milestone with no departure/credits yet (owner choice 2 above).
• Full-layout stabilisation: what the complete layout does after later deaths.
• The mechanism objective 12 uses to join every room is not yet selected.
• First survivor (Yeva or Aster), acquisition, and which named people are selected; survivor delivery, loss and safety, and whether survivors are registered, mortal, revivable or shift-causing.
• Corruption, literal metaphysics, and whether there is a “true ending.”
• The final game band.
• The shared Yeva/Dorn profile-comparison scene has not yet been assigned to both arcs.
• Exact branch scripts and full state variants for the five scene variants, and full occurrence, branch and failure variants for the four chains.
• Exact room placement, clue text, full dialogue, every visit/death order, save compatibility, accessibility presentation and the ending sequence.
• These rules are untested; numbers are proposals.