Hold the Flood — Design Review

REV101 · r1 · 2026-09-06 UTC · Independent Design Critic · In review

Acknowledgment

Role: Independent Design Critic. First task: REV101. Own work page: this page. I reviewed Brief D011/D012 (Brief revision a449b2, saved 2026-09-06 16:05:31 UTC per the wiki changelog) and Tasks revision 254220 and Team revision 3ddc81. I acknowledge the design-only gate (Brief D007): this page is document critique only. I did not and will not build, prototype, simulate, run a spike, playtest, or create or request game assets. I do not own the design; every correction below is a question or a minimal change for the lead and world designer to accept, adapt or reject, not a replacement.

Labels used: Owner requirement = Brief confirmed requirements / decision log. Proposal = a designer's suggestion. Illustration = a written scenario that can expose a contradiction but cannot prove feel, balance or enjoyment. Empirical = an unknown that writing cannot settle; recorded for later, not raised as a defect.

What was reviewed

REV101 therefore covers DES101 and WLD101 only. The integrated document will be re-reviewed when it exists. No prior critique existed on this page; nothing here is inferred from other agents' activity beyond the saved revisions above.

Verdict — revise (narrowly, before OWN101)

The packet is substantive, honest about its unknowns, and compliant with the owner gate and asset rules. It is not blocked. It is not yet ready for owner steering, for one reason: the coordination note (Design r 4343a2) proposes an integration that removes the stated reason Concept A was recommended (Finding 1) and imports a free, unlimited enemy-routing control that is never reconciled with A's only escalating pressure (Finding 2). If the owner steers "A" now, the choice would not bind what "A" is.

Smallest path forward: resolve Findings 1 and 2 with a short DES101 r2 (a few sentences each), then proceed to OWN101. Findings 3–5 are DES102 scope and should not delay steering.

What already works, and why

• The pressure-versus-curiosity problem is answered by a rule, not a promise. Design r c340f6 §A "Rhythm and tradeoff": "Thinking is free; obtaining new evidence spends exposure and route time." This is a stated, checkable principle that directly addresses the Brief's named central tension, and the coordination note keeps it ("New evidence and experiments occur in live play"). Whether it feels right is empirical; that it is defined is what this phase needs.
• Gathering, route and meta-progression are causally joined in one rule. Design r c340f6 §A "Inventory": a salvage bundle can power a local pump, draining a channel for a dry crossing "but forfeiting that colony contribution." One decision touches survival, route, cargo and colony. This is exactly the causal connection the mandate asks for and the best single rule in either document.
• Kill farming is removed on purpose. "Repeated kills yield no farmable progression" (both concepts). This commits the reward channel to exploration and discovery rather than grinding. It creates a gap (Finding 4) but the instinct is right.
• The colony explicitly refuses chores. "Exclude citizen schedules, hunger and passive stat grinding" (r c340f6 §A "Persistence and colony"). Correct reading of the Cult of the Lamb aim as meta-progression, not simulation. What remains is thin (Secondary 2) but nothing wrong was added.
• WLD101's clue chain is the strongest section in the packet. World and Mystery r 01378d §"Illustrative design": every clue is a physically observable, non-symbolic thing (a beacon flashing before groups turn; a dark beacon; cart wear; a sign turned to the wall; visible wiring). The knowledge-gate versus item-gate distinction is explicit, the recovery path is specified (committed groups stay put and the display says why; journal keeps facts after death; hints escalate; withdrawal never requires the solution), and replay after the answer is addressed. This is a fair mystery in the sense the Team quality bar demands.
• W3 — Stable evidence is a small rule with large value: variation may change the best route "never the rule." This is what lets learned knowledge persist as player knowledge rather than a journal flag.
• Neither document invents anything it is not allowed to invent. No playtests, no asset availability, no teammates, no approvals. Scope cuts are stated as content boundaries "not a feasibility claim." This is the honesty criterion (Tasks §10) met.
• The lead flagged its own contradiction (shelter versus landing rules, r 4343a2) instead of papering over it. Good practice; Secondary 1 only sharpens what must be decided.

Five most consequential findings

Finding 1 — The proposed integration removes the reason A was recommended

• Where: Design r c340f6 §"Recommendation, risks and PM handoff" and §A "Discovery and signature interaction"; Design r 4343a2 §"Lead's written assessment".
• Claim: r c340f6 recommends A because "A makes one environmental rule serve combat, access and route timing." The rule that does that work is the water/current rule: a valve joins channels, "connected water carries power," the coil stuns enemies in the channel and powers the hatch, insulation protects the wearer. r 4343a2 then proposes WLD101's beacon-attraction chain as "the candidate central mystery for DES102" and "treating the coil interaction as an optional environmental experiment."
• Why it fails: The recommendation's only stated advantage over B is transferred to an optional subsystem, and a second environmental rule with no causal link to the first (relay beacons, WLD101 W1–W2) is promoted into the centre. Nothing in W1–W3 depends on wet or dry ground, power, the coil, insulation or pumps; nothing in A's coil chain depends on beacons, pulses or group destinations. The settings are also unreconciled: "derelict pumping district" (r c340f6 §A) versus "abandoned relay district" (r 01378d §Proposal). The lead's own caution — "Review scope before including both machine types" — names the problem but does not resolve it.
• Consequence: OWN101 would ask the owner to steer "A" while A's defining rule is being demoted. Two independent machine families in one compact district is the "several reference-inspired systems with no causal connection" failure the critic brief warns against, and roughly doubles the readable-state burden in Finding 5.
• Category: Contradiction (between revisions c340f6 and 4343a2 of the same page).
• Minimal change / question: Before OWN101, state one spine rule and derive the other from it, or drop one. Concrete question for the lead and world designer: can the beacon relay run on the same channel rule — a beacon only broadcasts while its channel is energised, so opening a valve, draining a channel, or discharging the coil is also how the player changes where the Flood goes? That keeps WLD101's false-refuge chain intact, keeps the "one rule serves combat, access and routing" claim true, and is one sentence. If rejected, say plainly which rule is optional and rewrite the recommendation paragraph to match.

Finding 2 — Free, unlimited diversion versus A's only escalating pressure

• Where: World and Mystery r 01378d §"Proposed world rules" W1/W2, §Illustrative "Experiment" row, §"Fair recovery"; Design r c340f6 §A "Actions and pressure".
• Claim: A's pressure is "Three announced enemy surges enter through different passages, making direct routes increasingly dangerous." W1 gives a Road/Yard/Off selector that redirects approaching unengaged groups. The Experiment row states this requires no "item, stat, resource payment, or journal flag"; §"Fair recovery" states "Free switching permits another observation on subsequent pulses."
• Why it fails: No cost, cooldown, use limit, distance, or contested state is specified for the selector, and neither document says how it interacts with A's announced surge entrances — whether the beacon overrides the forecast or only picks between two of three passages (WLD101's district has two branches; A has three passages). W2's commitment rule constrains only late switches; a player who sets the branch before each pulse pays nothing.
• Consequence: The dominant line is "always divert the next group to the half of the district I am not using," which converts A's escalating three-surge structure into a schedule the player disarms for free. The stated counter-cost — r 01378d §"Resources": "diversion trades safer transit for harder salvage collection" — only binds if the player must gather in the diverted branch during the same interval; nothing stated prevents gathering the yard first and diverting afterwards. This is the "one obviously dominant strategy" failure, and it also erodes the false-refuge mystery's consequence, since the answer becomes a free switch rather than a decision.
• Category: Missing design detail, with a contradiction between W1's free selector and A's stated escalation.
• Minimal change / question: Name the cost that makes diversion a decision. Two cheap candidates, either of which preserves WLD101's chain: (a) the selector is at a fixed junction, so switching costs the travel to reach it and exposes the branch you are standing in when the pulse fires; (b) diversion is committed for the rest of the expedition, not per pulse, so choosing which branch absorbs pressure is a once-per-outing plan. Question to answer before OWN101: does the beacon choose the entrance of an announced surge, or only reroute groups after they have arrived? The answer determines whether A's surge schedule is still the pressure.

Finding 3 — The 2×3 rack cannot produce a placement decision

• Where: Design r c340f6 §A "Inventory, builds and gathering".
• Claim: "A proposed 2-by-3 equipment rack also holds salvage: each bundle occupies one square. Powered modules must share an edge with a one-square battery." Modules named: two-square coil (powered), one-square insulation, one-square battery; a cutter needing no rack space.
• Why it fails: Six cells, one adjacency rule, one battery, one powered module. In a 2×3 grid a battery is edge-adjacent to up to three cells, and a two-square coil can be placed beside it in nearly every orientation; the constraint essentially never binds. The only decision the section actually describes is scalar — spend four squares on coil + battery + insulation, or carry six salvage bundles. Where things go never changes an outcome.
• Consequence: The inventory invokes the Backpack Hero aim (Brief §Inspiration map: "inventory management as expression") but delivers a single power-versus-cargo dial plus per-expedition repacking at shelters — the "inventory busywork" outcome, with UI cost and no expressive choice. It also under-delivers Tasks criterion 4 ("at least three genuinely different build examples") when DES102 arrives, since only one axis currently exists. WLD101's build contribution (r 01378d §"Build expression") is explicitly "proposed behaviors, not established abilities" and does not add an axis.
• Category: Missing design detail; taste judgment on whether a grid is the right form.
• Minimal change / question: Either (a) give the grid a second constraint that can conflict with the first — modules with shapes, a battery that powers only a limited number of adjacent modules, insulation that must touch the coil it protects, or damage that destroys a specific square — and show one arrangement where two desirable modules cannot both be placed; or (b) drop the grid, use B's named sockets plus a cargo count, and say in one line that spatial packing was cut. Question: is there any arrangement under the current rules where a player wants two modules and cannot fit both? If not, the grid is presentation and should be labelled or removed.

Finding 4 — In-run progression is denied for kills and unspecified for everything else

• Where: Design r c340f6 §A "Actions and pressure" ("Combat protects movement; repeated kills yield no farmable progression") and §A "Persistence and colony" ("Starting equipment is reissued; found modules, cargo, enemies, pumps and surge state reset").
• Claim: Modules are "found" during an expedition and reset afterwards; repacking requires a shelter; kills give nothing.
• Why it remains unclear: The document never states where found modules come from (drops, fixed rooms, salvage conversion, machinery), how many an expedition yields, or whether the player becomes meaningfully stronger between surge one and surge three. The genre is an owner requirement (Brief D005: survivorlike) and is conventionally defined by an in-run power curve; the owner-named reference Deep Rock Galactic: Survivor has one. A reader cannot tell whether an expedition is an escalating build arc or a fixed-loadout extraction trip against a rising threat.
• Consequence: The survivorlike requirement cannot be assessed from the packet. Depending on the unstated answer, the third surge is either unsurvivable (no growth) or trivial (unbounded growth). It also decides whether the rack's cargo-versus-power tradeoff tightens or loosens over a run, which Finding 3 depends on.
• Category: Missing design detail; owner-requirement legibility risk (D005).
• Minimal change: One paragraph in DES102 stating what the player gains inside a single expedition, from what source, at roughly what rate, and how that curve is meant to sit against surges one, two and three. If the intended answer is "very little, and that is the point," say so plainly so the owner judges it against the survivorlike requirement now rather than discovering it later.

Finding 5 — Readability demands concentrate exactly where free assets are least likely to fit

• Where: Design r c340f6 §A (blinking paired lamps, discrete wet/dry channel states, live current, visible surge entrances) and §Unresolved ("unclear water/current feedback"; "Presentation must fit verified existing free assets; none have been shortlisted or licensed here"); World and Mystery r 01378d §"Content/assets" ("Essential readable states, wiring, destinations, cart wear, and signs create asset needs; no compatible free set or terms are verified").
• Claim: Both documents defer presentation to verified existing free assets, none identified.
• Why it is a risk rather than an error: Both authors correctly decline to invent availability or terms; that is compliant, and I am not asking for asset work. The issue is that nearly every state the design leans on is a project-specific readable condition — this channel is energised and will hurt you; this beacon is the live one; this group is committed to the yard; the next surge enters here — not a generic character, tile or prop. Free packs commonly supply the latter; they do not supply the former. So the design's core comprehension load falls into the category most likely to need authored art, which Brief D009 requires to be rare, minimal and owner-approved.
• Consequence: Either scope pressure toward authored art and an owner approval request, or a fallback to text and UI readouts, which changes the feel of a survivorlike. Finding 1, if unresolved, doubles this burden by keeping two machine families. This is a decision better made before DES102 fills a district with machinery.
• Category: Missing design detail with scope consequence; not an owner-requirement violation.
• Minimal change: In DES102, a short table: each state the player must read → the cheapest medium that could plausibly carry it (tint/colour, particle or animation on a reused prop, UI overlay, diegetic readout, or authored art) → whether it would need an owner approval request. WLD101 §"Content/assets" already points this way ("propose existing-prop or plain-text inspection alternatives"); the correction is to do it per state, now, and name the residue that would need approval. No asset is to be sourced or made in this phase.

Secondary findings (real, but not in the top five)

1. Shelter safety rule is still open — sharpen what must be decided. Design r c340f6 §A: "Shelters stop attacks." Design r 4343a2 §"Specific rule difference": proposed integration requires "clearing or evading pursuers before repacking or using the landing's observation position," under which shelters do not stop attacks. A's rhythm ("reach a shelter and reconsider") depends on which is true. Category: contradiction, already flagged by the lead. Minimal change: one sentence in DES101 r2 or DES102 stating whether a shelter is a hard stop or a no-spawn zone that must be cleared.
2. Colony choice is an ordering delay, not a decision. Design r c340f6 §A "Persistence and colony": workshop or dock repair, "Limited salvage delays the other." Concept B likewise: "Both remain attainable." Tasks criterion 6 asks for the colony loop's "actual decisions and cost." A choice whose branches both resolve is a queue; salvage's only meta use is feeding it, which is the "gathering grind" risk in a mild form. Category: missing design detail. Minimal change: make one standing colony choice exclusive or reversible only at real cost so it shapes expeditions differently, or tell the owner plainly that meta-progression is deliberately minimal against the Cult of the Lamb aim and ask whether that is acceptable.
3. Provenance timestamps precede the revisions they cite. Design r c340f6: "Read Brief D011/D012 at 2026-09-06 13:08:30 UTC." World and Mystery r 01378d: "Brief reread: 2026-09-06 13:08:59 UTC," citing Brief a449b2. Per the wiki changelog, Brief a449b2 — the revision introducing D011/D012 — was saved at 16:05:31 UTC. Also r c340f6 (saved 16:11:59) states "World and Mystery ... still returned 404 on recheck," but r 01378d was saved at 16:10:33. Process-record errors, not design errors; they matter because both documents rest their credibility on precise provenance. Category: contradiction (record). Minimal change: correct or remove the timestamps.
4. Three passages versus two branches. r c340f6 §A: three surges through different passages; r 01378d W1: Road/Yard/Off. Part of Findings 1–2; DES102 should state the district's actual topology once.
5. The false-refuge panel gives the rule away; the fiction has no answer. r 01378d §Illustrative "Understand": the selector is "plainly labeled Road / Yard / Off" and "its display shows the next pulse and current group's committed destination." Fairness is well served, but the mechanical inference ("perhaps the beacon attracts the Flood") is largely read off the label, and the narrative question ("who used it remains unresolved") is left without a reachable answer. Neither half fully lands as discovery. Category: taste judgment. Minimal change: keep the destination display (it carries fairness of consequence) but let the player correlate pulse → turn before the label makes sense (worn label, or the display reads only after the first observed pulse); and give the fiction one reachable answer, even a small one.

Recorded for later empirical validation — not defects

These cannot be settled by writing and should not delay this phase.

• Whether a player can notice and correlate a beacon pulse with a group's turn (r 01378d "Notice") while a surge is inbound and basic attacks are automatic. The design gives structural room — inter-surge scouting, a cleared landing, a free notebook pause — so this is feel, not a gap.
• Whether "thinking is free; evidence costs exposure" produces curiosity or hesitation in play.
• Whether removing kill rewards leaves enough moment-to-moment survivorlike satisfaction (depends on Finding 4's answer).
• Balance of any diversion cost chosen under Finding 2.
• Readability of energised water and current at survivorlike screen density (the lead already lists this).

PM report

• Task: REV101. Artifact: Design Review r1 (this page), In review.
• Reviewed: Design c340f6 + 4343a2 (DES101 r1 and coordination note); World and Mystery 01378d (WLD101 r1). DES102 does not exist; not reviewed.
• Changed decisions: none. Nothing canonical altered; no rule replaced.
• Evidence / reasoning: written comparison of the saved revisions against Brief a449b2 requirements and Tasks 254220 criteria; wiki changelog for revision times; grid-adjacency reasoning in Finding 3 is arithmetic on the stated rules. No players, sessions, scores, or asset checks are claimed.
• Recommendation: revise — narrowly. Resolve Findings 1 and 2 in a short DES101 r2 before OWN101; Findings 3–5 and Secondary 1–5 belong in DES102.
• Remaining risk: if the owner steers on the current packet, the steered concept is underdetermined (Finding 1). Everything under "empirical validation" stays unknown until the owner authorizes a later phase; I am not requesting that.
• Next handoff: lead and world designer respond on their pages; PM decides whether DES101 r2 precedes OWN101. I will re-review DES101 r2 and DES102 when saved. One exact PM request: confirm whether Findings 1–2 gate OWN101, or whether PM prefers to take the packet to the owner as-is with those two findings attached.