2026-09-06 18:57:35
Anonymous:
REV103 r1: present-tense review of DES103 r3 (Design Document r9, fb6b6d) on per-beat ledgers rebuilt from the rules; PMR002 1, 2, 4 met on the page, 3 met on World r16 but not yet lifted; verdict revise, one short pass: REV103-1 %A712 leftover sentence and two ledger cells, -2 collection rule disobeyed by both scenes, -3 %A711 Fix ending contradicts World r16 and %A74's two-cable model, -4 three stale sentences; REV102-13 resolved; REV102 r6 moved to Archive/REV102
hold the flood/design review.md ..
@@ 1,64 1,134 @@
# Hold the Flood — Design Review
-
**REV102 · r6 · 2026-09-06 15:52 UTC · Independent Design Critic · Accepted, ready verdict declined (PMR002); REV103 assigned, blocked on the revised DES103** — a closing correction to REV102's record, not a review of a new revision. PMR002 declined the "ready for owner review" verdicts this page gave at r2 and r3, and on the evidence it was right to. The contradictions it cites are verified below against the current document (r8) in the present tense, together with what r8 has already fixed and what REV103 will do differently. REV102 r5–r1 are on the [REV102 archive](/Hold%20The%20Flood/Design%20Review/Archive/REV102); REV101 r8–r1 on the [REV101 archive](/Hold%20The%20Flood/Design%20Review/Archive); every revision is in [page history](/Hold%20The%20Flood/Design%20Review/history) (REV102 r5 e2db8b · r4 041852 · r3 382607 · r2 82a96b · r1 4dde90 · REV101 r8 11bd4d · r7 bb6871 · r6 a89227 · r5 d3679b · r4 3fd6df · r3 698c13 · r2 47d51d · r1 7ad187).
+
**REV103 · r1 · 2026-09-06 16:05 UTC · Independent Design Critic · In review** — independent review of **DES103 r3, the Design Document at revision r9**, the revision written against PMR002 findings 1–4 and criteria 1–11. Present-tense, revision-specific, built on per-beat ledgers this role derived from the document's own rules before trusting its prose, as REV102 r6 committed to. REV102 is closed; its r6–r1 are on the [REV102 archive](/Hold%20The%20Flood/Design%20Review/Archive/REV102), REV101 r8–r1 on the [REV101 archive](/Hold%20The%20Flood/Design%20Review/Archive); every revision is in [page history](/Hold%20The%20Flood/Design%20Review/history) (REV102 r6 538724 · r5 e2db8b · r4 041852 · r3 382607 · r2 82a96b · r1 4dde90 · REV101 r8 11bd4d … r1 7ad187).
-
## What was checked
+
## What was reviewed
| Artifact | Revision | UTC | Note |
| --- | --- | --- | --- |
-
| **PM Reviews — PMR002**; Tasks; Hub | [PMR002](/Hold%20The%20Flood/Pm%20Reviews/Pmr002); 95db05; e05454 | 15:36–15:38 | Reviewed DES102 r6, WLD r13 and REV102 r3. Verdict: revise before PM recommends owner approval. REV102 accepted as delivered; its ready verdict declined. REV103 assigned to this role, blocked on the revised DES103 |
-
| **Design Document — DES103 r2 (document r8)** | d37cbf (r7 44984c) | 15:33 | Written before PMR002 and before REV102 r5; fixes REV102-10, -11, -12 and the cable vocabulary as EXT102 X21/X22 proposed; does not address PMR002 1–5 or REV102-13 |
-
| Brief 28fb1c, Team cbf8d7, World and Mystery e2d4b4 | unchanged | | |
+
| **Design Document — DES103 r3 (document r9)** | fb6b6d (r8 d37cbf) | 15:47 | Diffed against r8; §1, §4–§6, §8–§9, §11–§17 and both appendices read in full; §12 and §13 re-tallied from the rules |
+
| World and Mystery — WLD102 r15, r16 | dca569 → 812fd6 | 15:47, 15:49 | r15: what counts as a beacon; retarget-before-spread; state table of the signature routes; complete tier-three chain and endings table. r16: merges the lead's r9 access chain with the page's relay chain; two reconciliations stated. Canonical for W1–W4 and the chain |
+
| Tasks; PMR002 (closing) | ab97dc; 4f18c7 | 15:49, 15:48 | REV103 listed Blocked on "the revision addressing PMR002"; r9 is that revision and postdates the closing pass by two minutes. REV102-13 accepted by PM; quantity checks placed in this role's remit |
## Verdict — **revise: one short pass, then it can be judged for the owner**
-
REV102 r2 (on draft r3) and r3 (on draft r5) said the document was ready for owner review as the DES103 base. PMR002 finds three sets of contradictions that were present in r3 and r5, in sections this page said it had traced. Each is verified here against r8, where every one still stands:
+
r9 does what PMR002 asked, and the ledgers below show it: the scenes now obey the routing rule, the racks are legal and drawn, every unit of charge is tracked and the tallies close, the arc is drawn twice and played once, the fuel policy is stated as rules, the hours arithmetic is honest, the hauler's cost is true, the ending costs no salvage. PMR002 findings 1 and 2 are met on the page.
-
| PMR002 | The sentences, in r8 | Why it is a contradiction | This page's record |
-
| --- | --- | --- | --- |
-
| **1a** — the coil route in §1 | "The water carries the current to the hatch and it opens for a few seconds… **The yard stays dark.**" (§1, coil route) | §4 clause 2 puts the hatch and the yard beacon on one circuit; energised water lights the beacon steady for those seconds. §8 says so itself ("So does the yard beacon: it lights steady, because it is on the same wet circuit"). §1 and §8 disagree | Missed at r1–r5 |
-
| **1b** — the failed experiment in §8 | "Surge one, which follows the nearest lit beacon it can reach, **turns off the causeway and comes to the yard**" while the panel is still on causeway | The crowd is gathered *at* the lit causeway beacon; under "nearest lit beacon it can reach" its present beacon is nearer than the coil-lit yard. The scene's result is impossible under the rule it quotes. World r13's qualification ("any crowd nearer to B2 than to its own lit beacon") is the consistent version: the coil-lit yard draws stragglers and thinned organisms nearer to it, not the gathered crowd | Missed at r1–r5; the §8 lesson was checked, the movement claim was not |
-
| **1c** — precedence | "A surge whose beacon goes dark spreads along its branch at once, both ways" (§4), unconditioned; the plain-terms bullet two lines later has crowds walking to the new light | One sentence: retarget when another lit beacon is reachable; spread when none is | Not raised here |
-
| **1d** — the junction lamp | "crowds can be drawn toward the junction while its lamp is lit" (§8, §11) | Nothing says whether the pool's lamp is a beacon. If every lit lamp lures, say so and price it; if not, remove the claim | Not raised here |
-
| **2a** — §12's rack | "Rack now: coil, battery, three bundles, one slot empty" | Two plus one plus three is six of six; there is no empty slot. A seventh cell is claimed | Missed at r1–r5; the charge was traced, the slots were not counted |
-
| **2b** — §13's cargo | Starts as the electrician (`C C B / S S I`, two bundle slots), "gathers two bundles", ends "Extraction with three bundles" | Seven slots in a six-slot rack; the third bundle is never picked up | Missed at r1–r5 |
-
| **2c** — the hauler | "never opens a powered door" | The panel is free and opens one; the sacrifice is portable power and harvesting, not doors | Not raised here |
-
| **3** — hours | "ten to twelve expeditions… about fifteen minutes each… roughly three to four hours" (§11) | Ten to twelve times fifteen minutes is two and a half to three hours; the rest is unspecified overhead and should be labelled as such | Not raised here |
+
What remains is small and specific: one leftover sentence in §12 that contradicts its own correction; one new rule that both scenes disobey; §11's endings, which the world designer has since reconciled on the canonical page and which the document still states in the version that contradicts its own cable model; and three stale sentences PM has already asked to be removed or added. None is a rule change. This role does not pre-approve r10; it names what r9 lacks, and will judge r10 when it exists.
+
+
## The ledgers — expedition one (§12), rebuilt from the rules
+
+
Rules applied: six slots; battery 1, coil 2, bundle 1; a mote adds up to capacity ten; one unit a minute per bundle connected to a battery through touching bundles; modules go into empty slots wherever found; arranging and linking only in a shelter.
| ~4:00–5:00 | two stragglers → two motes; platform: `C C B / S - S`, link B→C; battery (1,3) touches coil and one bundle | arranged; one drip; stored 2 | 2, then −1/min | Yes — one drip is right for that drawing |
+
| 7:30 | third bundle into (2,2): `S S S` row, all connected to the battery = **6/6, three drips** | coil 2, battery 1, bundles 3; full; three drips; stored 0 | 0 | Yes — **but see REV103-1: the prose's next sentence says the opposite** |
+
| 9:00 | cutter kills two pursuers | not in the ledger | two motes dropped; collected or expired in seconds under three drips | **Ledger omits them** |
Timing note: with one drip from about 5:00, the two stored units are gone by about 7:00 on their own; the prose attributes their loss to the three drips at 7:30. Either the third bundle is added straight after arranging, or the ledger's 7:30 row starts from zero. One cell.
-
REV102 r5's own "revise" verdict on r7, for the arc, stands, but it was reached without seeing these; r2 and r3's "ready" verdicts are withdrawn. The verdicts were wrong for a specific reason that is correctable: this page traced charge and routing across the scenes and read slot counts, lit circuits and durations for plausibility instead of tallying them per beat. **For REV103 the method changes:** for each written expedition, a per-beat ledger — slots occupied and by what, links, drips, battery units, beacons lit and by what, each crowd's nearest lit beacon, door states, and elapsed time — built by this role from the rules before reading the prose, and the verdict given only against that ledger. That is the check that would have caught 1a, 1b, 2a and 2b, and it is in this role's mandate ("undefined rules hidden by confident language").
+
## The ledgers — expedition six (§13), rebuilt from the rules
-
## What r8 already fixes, verified
+
Batteries: coil's (B1) and arc's (B2), capacity ten each; collection "to the linked battery with the most room" as §5 states.
| 3:00 | two motes | **no link has been drawn yet** (the first arrangement is at 5:00), so under §5's sentence no battery is "linked" and the motes should expire | 2 / 0 | **Rule and scene disagree — REV103-2** |
+
| 4:00 | panel to yard; causeway crowd's beacon dark, yard the only lit beacon → retargets | — | same | Yes — precedence obeyed |
+
| 5:00 | + arc = 6/6; harvester `C C B / A B I`; B1→C, B2→A; B1 at (1,3) touches C and I; B2 at (2,2) touches C, A, I — no bundle, no leak | 2 / 0 | 2 / 0 | Yes |
+
| 6:00–7:30 | eleven kills → eleven motes; by "most room" B2 (room 10) fills before B1 (room 8) until they equalise, so about 6 / 5 | — | 10 / 3 (B1 first, to full) | **Scene fills the fuller battery first — REV103-2** |
+
| 7:50 | discharge B1 → 0; 29 stunned → two large motes = 10; B1 has the most room | 10 / 3 | 10 / 3; spent 10 | Yes |
+
| 8:00 | discharge B1 → 0 into ~60 fresh in water → six motes = 30; B1 takes 10, B2 takes 7 (3→10), 13 expire | 10 / 10; expired 13 | same | Yes |
+
| 8:20 | burst: B2 10 → 0 over ten seconds; targets have shed → no motes | 10 / 0 | same | Yes |
+
| 8:30–11:30 | causeway empty (both crowds at the yard); junction gate shut; valve opened → pool lamp (indicator) lights, gate opens; player does not enter | 10 / 0 | same | Yes — consistent with §2, §4, §11 |
+
| 12:00 | causeway home; no cargo; B1's ten lost at the dock | — | same | Yes |
+
+
Both scenes now close to the unit. The two disagreements are between the scenes and one new sentence in §5, not within the scenes.
+
+
## PMR002 findings 1–4 against r9
+
+
| Finding | Where in r9 | Status |
| --- | --- | --- |
-
| REV102-10 — "fills a battery" | §4 clause 4: "adds its charge to a battery in your rack, up to the battery's capacity; with no battery, or a full one, the charge is lost" | **Resolved** |
-
| REV102-11 — the electrician's loop | §6: "With one battery it breaks even… The electrician who wants a profit carries a second battery in place of a bundle… the printed rack in section 5 is the break-even version" | **Resolved**; PMR002 2's note that the two-battery layout also needs leak accounting applies to it |
-
| REV102-12 — the arc's bound | §6: the drain is now a full battery in about ten seconds, "set above what the cutter alone can earn"; the bound is restated as the surge (once-per-organism) | **Resolved as stated**; PMR002 2 asks further for the fuel policy — whether an active battery can be recharged mid-burst — which r8 does not state |
-
| REV102-7 — one cable vocabulary | "cable" throughout; one Appendix A entry | **Resolved** |
-
| REV102-13 — the arc undrawn, unplayed, cost unstated | No loadout carries an A; neither scene uses it; no §14 row | **Open** — PMR002 2's "specify the arc's fuel policy and show early/middle/late changes" is the same requirement from PM's side |
+
| **1** — signature experiment vs the routing rule | §1 coil route: "The yard's beacon, on the same circuit, glows steady for those seconds… but it never flashes"; §1 state table (three routes: panel, coil, switch-then-coil, each with panel, circuit, lit beacons, crowd position→target, hatch); §4: "Precedence, stated once: a crowd retargets to the nearest lit beacon it can reach whenever one exists, and spreads… only when none does; a crowd already holding at a lit beacon never leaves it for a farther light. Only beacons pull; other lamps, including the junction pool's indicator, are not beacons"; §8 rewritten so the failure needs a crowd with nowhere to go (panel off, surge from the sluice spreading) — and says why the old version could not happen. World r15's state table reaches the same rows | **Met.** I checked each state-table row against clause 3 and the precedence; all four follow |
+
| **2** — space and charge | Legal drawn racks in §5 (three loadouts, the failed arrangement, two arc loadouts with adjacency stated and no leaks — I checked each battery's neighbours); charge's five states; collection, overflow, planting and retrieval rules; the arc's fuel policy (one battery per burst; no charging while drawing; no motes from shed organisms); early/middle/late; §12 and §13 ledgers | **Met, with REV103-1 and -2** |
+
| **3** — final mystery and endings | §11 tier three: a complete clue → inference → action → consequence chain on the existing rule; each ending's dispatch, forecast, retreat and door consequences; the ending costs no salvage; hours restated honestly; early-stop answer without achievement figures. World r16 merges this chain with the page's relay chain and reconciles two points | **Met on the world page; the document lags it — REV103-3** |
+
| **4** — market evidence | §11 and §17: measures named with limits; no minimum-length or refund conclusions; the "minority who finish" forecast removed; the intended player named once; price uncommitted | **Met** (REF103's own corrections are the researcher's) |
+
+
## The findings
+
+
### REV103-1 — §12 keeps the sentence PMR002 2a found, two sentences after correcting it
+
+
- **Where:** §12, 3:30–7:30. The new text: the platform arrangement `C C B / S - S` with one drip; then "the player finds a third deposit… drops the bundle into the empty slot. Now the three bundles touch each other in a row, the row touches the battery, and three drips show; the two units are gone in under a minute." Then, unchanged from r8: "**Rack now: coil, battery, three bundles, one slot empty, and one drip, because a bundle touches the battery.** The player does not yet know what a drip means."
+
- **Why it fails:** six of six slots are full and three drips show, by the paragraph's own previous sentence and by the ledger's 7:30 row. The stale sentence asserts a seventh slot and one drip — the exact defect PMR002 2a named, now contradicted by its own fix.
+
- **Also in the ledger:** the two pursuers killed at 9:00 drop two motes the ledger does not record; and with one drip from 5:00 the two stored units are gone by about 7:00, before the three-drip moment the prose blames.
- **Minimal change:** delete the stale sentence and its duplicate "does not yet know" clause; add "dropped 2; expired 2" at 9:00; either place the third bundle immediately after arranging or start the 7:30 row from zero.
+
+
### REV103-2 — The new collection rule is disobeyed by both scenes
+
+
- **Where:** §5: "A collected mote goes to **the linked battery with the most room**." §12 ~4:00: the battery collects two motes before the first arrangement at the platform — it is not yet linked. §13 3:00: the first battery collects two motes; links are drawn at 5:00. §13 6:00–7:30: eleven motes go "eight to the coil's battery, which is now full, and three to the arc's" — the battery with *less* room fills first.
+
- **Why it fails:** the sentence was added in r9 to answer PMR002 2 and it is contradicted by both scenes that were rewritten in the same revision. Either unlinked batteries collect (the scenes) or they do not (the rule), and either allocation is by room (the rule) or by the player's choice or by the coil's battery first (the scene).
+
- **Consequence:** small in play, but it is a rule about the arc's fuel, which the recommendation rests on; PMR002 asked for the fuel policy as rules and this is the one rule in it the examples break.
+
- **Category:** contradiction (rule versus its own examples).
+
- **Minimal change:** one sentence, either way. Simplest, matching the scenes: "A collected mote goes to the battery with the most room, linked or not; a full battery takes nothing; a battery the arc is drawing takes nothing until the burst ends" — and make the 6:00 split follow it (about six and five), or state that the player picks the battery when both have room.
+
+
### REV103-3 — §11's endings contradict the canonical chain and the document's own cable model
+
+
- **Where:** §11 tier three, "Fix it": "the panel is welded to one branch; that branch's beacon flashes at every dispatch for good… and **the welded branch's door is open forever while the other's is shut forever**." World and Mystery r16 (canonical for the chain, by the two pages' own division of sources) merges the lead's chain and reconciles this: "**Fix** clamps the signal cable only, so the intake is fixed and power still switches, and the lead's 'welded branch's door open forever, the other's shut forever' is dropped, because under the two-cable rule a clamp on the lever cannot fix the power line."
+
- **Why it fails:** §4 makes power and signal two cables switched separately; a clamp on the lever fixes what the lever does. The document's Fix ending has the clamp fix both, which §4 forbids; the world page's version does not. The document also lacks the relay chain's inside steps (drum, lever, clamp, drain valve) that give the three endings their observable cause. The lead offered the chain "for the world designer to correct and own"; the correction now exists and has not flowed back.
+
- **Consequence:** the finale — the thing PMR002 3 asked to be complete — is stated two ways on the two pages, and the document's way breaks the document's own rule.
+
- **Category:** contradiction (cross-page, and with §4).
+
- **Minimal change:** lift World r16's "Reaching the room", "Inside the room" and the endings table into §11 verbatim, as §9 does for tier one; keep Shut's fifteen-minute window and hard-stop docks (r16 already adopts the lead's version of that).
+
+
### REV103-4 — Three stale sentences PM has already asked about
+
+
- Owner page: "An independent critic has reviewed this document and found it consistent, honest about its unknowns, and readable cold" — that was true of r5 and misrepresents the current state; PMR002 5 asks for "reviewed; revisions remain open". Say that.
+
- §16: "A critic outside the recorded team has posted a market and originality audit; PM has not yet ruled on its standing" — the External Critic is a team member since D015, and the paragraph's substance is now in §17. Remove or date it.
+
- Appendix A has no **battery** entry, the one Q39 asked for and PMR002 assigned; the word is used in every section.
+
+
**Category:** housekeeping; criterion 10.
+
+
## What works in r9, and why
+
+
- **The scenes are now state sequences**, and they obey the rule they quote. §1's table is the first place the document lets a reader *check* the signature interaction rather than believe it; §8's rewritten failure is a better failure than the old one, because it needs the player to have made a crowd with nowhere to go.
+
- **The ledgers close.** Expedition six tallies to the unit through two discharges, an overflow of thirteen, and a burst that returns nothing because its targets have shed — the fuel policy demonstrated, not asserted. That is what PMR002 2 and REV102-13 asked for.
+
- **The arc costs what the link rule says it costs**: six slots and no cargo for the harvester, or three bundles and a slow burst for the arc hauler; both drawn, both leak-free by adjacency I checked.
+
- **The tier-three chain uses only what the player already knows** — the junction valve's lesson, one pool further — and the world page's relay chain gives each ending an observable cause in the same vocabulary. Once §11 lifts it, criterion 5 holds at the finale as it does at tier one.
+
- **The market text now says what the measures are**, and the length is two figures with their assumptions labelled.
-
## What REV103 will review, when the revised DES103 exists
+
## Recorded for later empirical validation — not defects
-
PMR002 findings 1–4 and criteria 1–11, on the exact revision, including unchanged surrounding examples, with the per-beat ledgers above; REV102-13; the hours label; the junction lamp's status; the tier-three chain and each ending's consequences (World's, checked in the document). A present-tense verdict with evidence; no pre-approval. Nothing is posted before that revision exists except a PM correction.
+
§16's list, unchanged; the arc's actual duration with pickups (§6 states the nominal); the press-rate bound.
- **Task:** REV102 closed as delivered; ready verdicts withdrawn; REV103 acknowledged, blocked on the revised DES103. **Artifact:** Design Review, REV102 r6 (this page); r5 moved to the REV102 archive verbatim.
-
- **Checked:** PMR002 in full; Tasks 95db05; Design Document d37cbf against PMR002 1–3 and REV102-10 to -13, by grep and by re-reading the cited sentences.
-
- **Changed decisions:** none by this role; two of its own verdicts withdrawn.
-
- **Evidence:** the r8 sentences quoted above; the slot arithmetic; the hours arithmetic. No players, sessions, scores or asset checks; nothing built, prototyped, tested or created.
-
- **Remaining risk:** that REV103 repeats the method that missed these; the ledger discipline above is the mitigation, and PM can hold this role to it.
-
- **Next handoff:** lead and world designer — DES103 r3 against PMR002 1–5 and REV102-13; then REV103 on that exact revision. **One exact PM request:** none; the assignment is clear.
| REV102-1 to -12 | REV102 | Resolved at DES102 r3 – DES103 r2 |
+
| REV102-13 — the arc: undrawn, unplayed, cost unstated | REV102 r5 | **Resolved at DES103 r3 (r9)**: two drawn loadouts, fuel policy, §13 burst, §14 row |
+
| PMR002 1 (signature/routing), 2 (space and charge), 4 (market wording) | PM | **Met in r9** (2 with REV103-1, -2) |
+
| PMR002 3 (finale) | PM | Met on World r16; document lags — REV103-3 |
+
| **REV103-1** — §12's leftover sentence and two ledger cells | REV103 r1 | **Open** |
+
| **REV103-2** — collection rule vs both scenes | REV103 r1 | **Open** |
+
| **REV103-3** — §11 endings vs World r16 and §4 | REV103 r1 | **Open** |
+
| **REV103-4** — three stale sentences | REV103 r1 | **Open** |
+
+
## PM report — REV103 r1
+
+
- **Task:** REV103. **Artifact:** Design Review, REV103 r1 (this page); REV102 r6 moved to the REV102 archive verbatim.
+
- **Reviewed:** Design Document fb6b6d in full; World and Mystery 812fd6; PMR002 closing and Tasks ab97dc. Both expeditions re-tallied slot by slot and unit by unit from §5's rules; every drawn rack's adjacency checked; each §1 state-table row checked against clause 3 and the precedence.
+
- **Changed decisions:** none by this role.
+
- **Evidence:** the two ledger tables above; the quoted sentences. No players, sessions, scores or asset checks; nothing built, prototyped, tested or created; no external source verified or repeated as fact.
+
- **Verdict:** revise — one short pass (REV103-1 to -4: one deletion, one sentence, one lift from the world page, three housekeeping lines). r9 meets PMR002 1, 2 and 4 on the page; 3 is met on the world page and awaits the lift. No verdict is offered on r10 before it exists.
+
- **Remaining risk:** the two pages diverging again on the finale if §11 is rewritten rather than lifted; the ownership rule in r3/r12 covers it.
+
- **Next handoff:** lead — r10 with REV103-1 to -4, lifting World r16's chain and endings verbatim; PM — the full review PMR002 says later revisions require. I review r10 when it exists. **One exact PM request:** none; note that REV103 is unblocked by r9 and this is its first present-tense review.
