Hold The Horde — Production Scope
The game uses Unity and Steam and favours free existing assets, as the Owner Brief requires. The SMALL band below is a costing sample, not a recommendation to replace the owner's ambition with a short game. Its purpose is to expose what we must actually author and what repeating a room does not save.
What the inventory counts
The Whole Game Outline offers three bands. Keep reusable definitions, placed instances, investigations and scenes separate. A discovery objective can use an existing room, trap, survivor and language chain; that intersection is not five additional encounters.
| Content unit | SMALL sample | TARGET / STRETCH | Work that reuse does not remove |
|---|---|---|---|
| Regions | 4 | 7 / 10 | A regional purpose, entry/return opportunities, pressure and discovery progression |
| Reusable room modules | 24 | 45 / 70 | Readable internal geometry, accessible controls, fixture and entrance variants |
| Placed playable room instances | 40 | 80 / 130 | Stable identity, actual contents, valid interfaces and recovery in its permitted arrangements |
| Expedition objectives | 12 | 24 / 40 | Entry condition, action, consequence, alternate access and later use |
| Ship-repair assemblies | 4 | 7 / 10 | Acquisition/installation consequences |
| Survivors and scenes | 5 people, 15 scenes | 9/45; 14/98 | Distinct motives, short rescue, home follow-through and limited knowledge |
| Equipment definitions | 12 | 24 / 36 | Source, reusable behaviour, limitations and meaningful combinations; the first twelve briefs already exist |
| Ordinary / special enemy roles | 6 / 3 | 10/6; 14/9 | Distinct pressures, readable responses and combined encounters |
| Major set-pieces | 2 | 3 / 5 | Exceptional staging and recovery; hosted rooms/objectives remain the same units |
| Mechanism families / major placed mechanisms | 5 / 14 | 9/24; 14/40 | Fourteen proposed major instances need geometry, warnings, stages, persistence and rearming; fixed room services and ordinary joins remain separately owned interfaces |
| Language chains across at least two languages | 4 | 8 / 12 | Evidence, alternatives, discriminating action, correction and useful transfer |
| Mystery or personal investigation chains | 4 | 7 / 11 | Causal evidence and consequences rather than additional exposition |
| Person-linked home projects | 5 | 9 / 14 | Voluntary contribution, an observable result, reclaim and personal consequence |
| Special connection challenges | 2 | 4 / 6 | Constraint/recovery problems, not the count of ordinary joins |
The fourteen proposed major mechanisms are not every interactive object: R26/R33 service transformations, the R02/R03 board/rack, ordinary doors and join faces stay owned by their room/interface descriptions rather than becoming extra P-identities. Campaign Evidence And State Crosswalk contains 26 index entries across the existing eight chains and fifteen scenes. They are deduplicated index rows, not 26 new objects or single captures.
The current count includes P14, the moving scan and bridge: substantial R15 rail/scan-bay geometry, a glazed bridge opening and span, a transfer-void sightline, an R28 receiver and shared E05/E07 source, trigger and revisit state. It uses existing rooms and encounters without adding a source volume. Its effect on the allowance below remains uncosted.
The 15 SMALL survivor scenes include rescue and home follow-through; home-project scenes should be assigned within those fifteen where possible. Branches are variations of a named scene, but each materially different consequence still needs writing and state review.
Likewise, a language chain and a mystery chain may share evidence and an objective. Give the shared observation one content identity, then separately specify its linguistic inference and narrative consequence. A count of eight chains does not imply eight disjoint routes or eight extra objective sets. Conversely, calling two investigations the same chain cannot erase a distinct test or correction.
Forty instances from twenty-four modules means sixteen placements beyond one of each module. It does not mean sixteen free rooms. Each placement must earn its reuse with a different useful relationship, pressure, consequence or meaning; colour changes and new neighbours alone do not establish discovery depth. The all-station puzzle includes all forty playable instances under this sample. The two special challenges do not cap its ordinary joins.
The TARGET and STRETCH columns are alternatives, not extra counts to add to SMALL. Increasing instance or survivor counts cannot be converted directly into author-hours: branch interactions, clue dependencies and permitted arrangements can change the work.
A shared unit ledger
Before detailed authoring, every unit needs an owner, definition or instance identity, current status, dependencies, reuse links and one owning description. Design v1 needs a ledger and behavioural briefs; the later campaign specification fills in each placed instance, clue and scene through its actual state changes.
The outline's pressure-ledger expedition illustrates the accounting:
| Shared unit | Own once | Additional work belongs to |
|---|---|---|
| Sampling ward and its installed Impact Dynamo | One placed room and one part instance, using existing module/part definitions where appropriate | Placement specifies removal consequences; equipment specifies Dynamo behaviour |
| Valve sample/purge test and visitor destination evidence | One observation/action sequence, referenced by relevant objective and chains | Language specifies competing interpretations and correction; story specifies what the transfer evidence does and does not establish |
| Yeva's return conversation | One named cast scene, potentially one of the three in her arc | Cast specifies changed belief and dialogue intent; home specifies only any distinct project operation |
| Regional horde and mechanism | One hosted encounter with references to role/family definitions | Placement specifies geometry and timings; roster defines the creatures, information design defines presentation |
| A recorded result across death | The same evidence and world instances | State contract owns continuity; information design owns how historic evidence is displayed |
Actual objective links belong in the Regional Discovery Map.
Three depths of completion
Complete wiki v1: every major domain has a usable rule, whether or not the owner has chosen it; alternatives with consequences are decided or explicitly left open; and the inventory explains progression, rewards, cast, both languages, endings and interfaces. There must be enough concrete behaviour to expose missing links. A domain containing only a title or “TBD” fails this milestone.
V1's room/mechanism inventory must contain actual briefs, not only count allocation:
| Unit | Minimum v1 output | Additional detailed-campaign work |
|---|---|---|
| Each of 24 reusable modules | Internal geometry, entrance/fixture relationships, available actions and intended discovery/pressure role | Full variants and compatible interface combinations, detailed cue/interaction placement and exceptional states |
| Each of 40 placed rooms | Named identity/module, region and objective links, actual contents/reward, first-visit/revisit purpose, access and recovery constraints | Exact per-instance placement, every selected branch/transient state and concrete consequences across permitted arrangements |
| Each of 14 proposed major placed mechanisms | First-version geometry, warning, action/stages, reset/rearm and persistence boundary, usable alternative/recovery and linked discovery | Full variant/combined-encounter details, exact thresholds and exhaustive state/recovery paths |
A room brief may reference its module and a mechanism brief, but cannot omit its own distinguishing content.
Detailed campaign specification: every approved placed room, mechanism, clue, scene, reward and ending branch has concrete entry/exit, observation, action, interruption, persistence and recovery behaviour. Shared definitions are referenced; instance-specific differences are written out. This is larger than a map of twelve objectives.
Production and validation: engine architecture, actual tools, runtime content, assets, integration, performance, accessibility testing and real-player evidence. This is a separate phase, not yet authorised.
Detailed-campaign scope and allowance
Detailed campaign work extends first-pass briefs into exhaustive content; it does not count their existing definitions again. The original allowances below cover deeper documentation for the SMALL sample with 13 major mechanisms, excluding the current first pass. They are planning estimates, not a schedule. The effect of P14 and revised placement geometry on these allowances remains uncosted: 350–590 role-hours is the original allowance, not an updated total or ceiling for the current proposal. Shared observations, scenes and fixtures remain referenced once:
| Workstream | Unique deeper work, excluding current first pass | Judgment allowance |
|---|---|---|
| Rooms, objectives and arrangements | Exact stateful variants for 24 modules/40 rooms/12 objectives, permitted arrangements, traversal, recovery sides and the two special connection challenges. | 90–150 role-hours |
| Encounters and mechanisms | Original basis: exact geometry/population/cue/timing/rearm combinations for 10 encounters including 2 setpieces and 13 major mechanisms; fixed services stay in room work. | 55–95 |
| Campaign evidence and choice | Full clue distribution, corrections and consequences for 4 language + 4 story chains, deduplicated against objectives and rooms. | 50–85 |
| Cast, rescue and home | Final writing/state variants for 5 people/15 primary scenes/5 home projects; host and branch variants cost work without becoming new primary scenes. | 45–75 |
| System detail and information | Exhaustive equipment/survival/build edge cases, exact cue/string/control variants and accessibility equivalence, deduplicated against encounter/room states. Survivor dialogue is counted under Cast. | 65–110 |
| Persistence, compatibility and integration | Complete transition/reference/compatibility matrix, version policy, cross-domain review and corrections. | 45–75 |
| Original SMALL detailed-campaign total | Documentation and campaign content only; no implementation, assets, recruitment, player sessions or technical validation. | 350–590 role-hours |
The range is broad because actual branch density, permitted arrangements and final dialogue depth are not yet known. The evidence index has 26 rows but does not enumerate every occurrence, branch or failure variant. Three representative detailed packages—a repeatable mechanism room, a two-language investigation with correction, and a short rescue/home arc—would need to be written and independently checked before narrowing the range or extrapolating campaign duration. TARGET and STRETCH are alternatives whose branch interactions require their own estimate rather than multiplication from SMALL.
Production constraints to resolve
| Area | Why it can dominate cost | Design preparation now |
|---|---|---|
| Modular station and interfaces | Reused room art still needs compatible door faces, controllable mechanisms, readable traversal and recovery | A module/instance ledger and entrance/fixture requirements |
| Connected equipment and hordes | Combinations, targeting and dense pressures may interact beyond their individual briefs | Shared state/trigger contract, representative combinations and later technical questions |
| Saves and content changes | Persistent joins, items, notes and people must survive changes consistently | Stable identities and compatibility decisions from Save And Content State |
| Cast and clues | Branch knowledge, testimony and inference can contradict each other without any new asset | One evidence ledger with speaker knowledge and consequences, reviewed across chains |
| Visual/audio/UI coherence | Free assets may not fit one another or communicate the required mechanics/accessibility | Category requirements and substitute plans; inspect actual rights and suitability only when candidates are selected |
| Unity/Steam delivery | Engine/platform selection does not specify target hardware, input coverage, distribution or performance budget | Record those decisions and verify current primary documentation when proposing concrete requirements |
Likely asset categories are modular station surfaces/doors, machinery, astronaut/creature/survivor models and animation, UI typography/icons, sound/ambience and effects. No AI-generated art, including placeholders or concepts. A rare custom requirement or purchase must be made concrete for owner approval. Art availability must not be used to invent a licence, silently commission work or make a clue depend on unreadable decoration.
Two cost-cut proposals to compare
Concentrate the SMALL station: keep its four regions, twelve objectives, two languages, five survivors and twelve equipment definitions, but compare eighteen modules/thirty placed rooms instead of twenty-four/forty. The removed six definitions and ten placements must be identified in the future ledger; move or combine their actual discoveries without erasing tests/corrections. The entire-station puzzle still includes every remaining room. The saving is fewer layouts and placements to author, while the cost is less spatial variety and a smaller physical assembly.
Reduce staging variety: retain SMALL's rooms/objectives and five mechanism families, but compare six ordinary/two special enemy roles and one exceptional set-piece instead of six/three/two. Ordinary hordes remain meaningful, and distinctive discovery consequences must be rehosted rather than deleted invisibly. This saves one role definition and one bespoke sequence; it costs spectacle and pressure variety and may overwork the remaining encounters.
These are separate alternatives; adopting both requires evaluating their combined loss of variety. Neither adds oxygen grind, repeated death requirements, loot loss or a longer escort to compensate for fewer units. Preserve all owner pillars first, and judge the actual retained experiences before adopting a cut. Larger counts remain plausible if later authoring and player evidence support them.
Reopening conditions: a complete ledger reveals missing scenes/branches; a core rule changes many interfaces; owner scope or ending choices change; representative detailed packages exceed allowances; staffing fails to materialise; or later evidence shows repetition, poor comprehension or unaffordable runtime/content work. Any revised scope or forecast belongs in Progress.
Open questions
- The owner has not selected a game size (SMALL, TARGET or STRETCH), a production budget, any asset or purchase, or a development phase.
- Whether the owner's 30/90/100–200-hour ambitions can be reached is unproven; none of the counts establishes hours of play.
- The optional departure premise behind the ship-repair assemblies is unselected.
- Whether every permitted arrangement works, and whether a valid full assembly of all forty rooms exists, has not been shown.
- The Unity/Steam choice does not settle target hardware, input coverage, distribution or performance budget.
- Whether suitable free assets exist for each category is unknown.
- The hour saving and retained campaign duration of the station-concentration cut have not been estimated.
- These rules are untested; numbers are proposals.