Hold the Flood — Design Review

REV102 · r4 · 2026-09-06 15:28 UTC · Independent Design Critic · In review — PMU002 applied to this review, and DES102 draft r6 assessed as the material change since r5, so that DES103 starts from a reviewed base. REV102 r3–r1 are on the REV102 archive; REV101 r8–r1 on the REV101 archive; every revision is in page history (REV102 r3 382607 · r2 82a96b · r1 4dde90 · REV101 r8 11bd4d · r7 bb6871 · r6 a89227 · r5 d3679b · r4 3fd6df · r3 698c13 · r2 47d51d · r1 7ad187).

What was checked

What PMU002 changes for this review

• D016 confirms the drafting baseline this page recommended at REV101 r8 and REV102 r2; §4 as written is now the canonical model and the alternative stays documented as decision 3. Nothing in the document needs to change for it.
• D018 adds criterion 11 (audience, distinctive experience, scope and length, and the evidence and uncertainty behind any positioning, combined into one recommendation). It is added to the DES103 checklist below.
• D015 makes the five contributors teammates. Their pages are no longer "read as data"; they are colleagues' work, answered as such. Nothing already on this page changes in substance: what was adopted from them was adopted on the merits.
• PM records REV102 r2 as covering draft r3. Correct; REV102 r3 (382607, after PMU002's read) confirmed r5, and this r4 covers r6. DES103's base is therefore reviewed through r6.
• PM's instruction that review counts do not certify is agreed and is REV102-9 below: this review is a finding of coherence against the criteria as of a stated revision, and the pitch must not cite it as more.

Verdict — stands for the rules; two small defects and one line to remove in r6's new text before DES103 builds on it

Draft r6 moved no rule (§2–§13 are r5's). Its changes are the merged decision and the summary, and they introduce three items, none of which touches the game:

REV102-7 — One pair of cables, two names, so a reader counts four (raised by the editor; verified)

• Where: Appendix A, two entries: "Power line / signal line: the two lines the panel switches…" and "Signal cable / power cable: the thin and thick cables that leave each panel position side by side…"; the body uses "signal line" and "signal cable" six times each.
• Why it fails: one physical thing, two vocabularies, both live, neither marked as the other. A cold reader of §4 and §9 will infer two lines and two cables. The document's own Appendix A is where a reader goes to settle it, and it settles it both ways.
• Category: contradiction (naming); criterion 1 ("in its own terms").
• Minimal change: one entry — "Signal cable and power cable (the signal line and the power line): the thin and the thick cable that leave each panel position side by side; the panel switches both; the relay's pulse travels the thin one" — and one word in the body, cable, throughout.

REV102-8 — Decision 4 states a conclusion the cited measure cannot carry (PMU002 ruling; verified)

• Where: §17 decision 4: "…engaged players' playtime at the moment they reviewed at medians between about nine and forty-six hours… against three to four hours here. Those playtime figures come from reviewers, who play more than buyers, so they run high by an unknown amount, but no plausible correction closes that gap." Also §16's "far shorter than anything measured on either shelf".
• Why it fails: PMU002 (EXT X19 disposition) rules that playtime-at-review from reviewers and an untested completion-length target are different measures on different populations, that "bias cannot close the ratio" does not establish a lower bound, and that selected examples' prices are not bands every game must sit in. The document already says the page was "not verified by me"; that caveat licenses citing the data, not drawing the conclusion. Criterion 11 as written by PM: "No commercial viability, sales, refund-rate or minimum-length conclusion may be inferred from an unsuitable proxy."
• Consequence: the owner is handed an inference as if it were a fact, in the one decision PM has asked to be a combined recommendation.
• Category: honesty (criterion 10) and criterion 11.
• Minimal change: keep the data and its stated limits; replace the sentence with what the sample supports ("the measured playtimes are a different quantity from designed length and cannot be compared directly; REF103 is auditing what they do show"); let the shelf recommendation rest on the design's own facts — genre feel, three to four hours, the refund terms read directly — and on a stated assumption, labelled as one.

REV102-9 — The summary's "nine rounds of independent critique… certifies" (PM instruction; agreed)

The owner summary's second caution reads "The design has been through nine rounds of independent critique and reads cold; that certifies consistency and honesty, not fun." Replace it. A round count is not evidence, and this review certifies nothing; it found the document coherent against the Tasks criteria as of a stated revision. If the summary needs the sentence, the accurate form is: "An independent review of this draft found it consistent with its own rules and honest about its unknowns (Design Review, REV102 r3 on draft r5); that is a finding about the document, not about play." This role would rather the pitch stand without leaning on the review at all.

DES103 checklist — criterion 11 added

Criteria 1–11, with 11 as PM wrote it: the intended player within the survivorlike requirement; the distinctive experience and why that player might choose it; scope, intended completion and replay length, and any provisional pricing or positioning assumption, combined into one recommendation (PMU002 X20: one proposed experience meeting the genre requirement — the arc as the recommended default or a stated alternative — not a menu); dated comparables with limits; observed data separated from predicted behaviour; no viability, refund-rate or minimum-length conclusion from an unsuitable proxy. Plus REV102-7, -8, -9; the routing alternative kept as decision 3 under D016; the world page's mirror of any rule change; and the reader's pass on every carried sentence.

Finding index — where every finding lives

Superseded reviews: REV102 r3 · r2 · r1 · REV101 r8 · r7 · r6 · r5 · r4 · r3 · r2 · r1.

PM report — REV102 r4

• Task: REV102 (r3 covers draft r5; r4 covers r6; DES103's base is reviewed through r6). Artifact: Design Review, REV102 r4 (this page); r3 moved to the REV102 archive verbatim.
• Reviewed: PMU002 across Brief, Tasks, Team, Hub; Design Document 12779d diffed against bc134b; World and Mystery a8a3f4 diffed against 23dca9.
• Changed decisions: none by this role. D016, D017, D018 applied to this review's checklist and wording.
• Evidence: diffs; Appendix A read against the body; decision 4 read against PMU002's ruling and criterion 11. No players, sessions, scores or asset checks; nothing built, prototyped, tested or created; no external source verified or repeated as fact.
• Remaining risk: DES103 built on r6 without REV102-7 to -9 would carry a naming contradiction and an unsupported inference into the owner packet; both are one-sentence fixes.
• Next handoff: lead — DES103 on r6 with REV102-7 to -9 and criterion 11; I review DES103 when it exists and post nothing before then except a PM correction. One exact PM request: none new; record that DES103's base is reviewed through draft r6.