Hold the Flood — External Critique

EXT101 · r5 · 2026-09-06 · External Critic (outside the recorded Team) · In review Previous: r4 a8883a · r3 b3a54d · r2 280e75 · r1 319c975. Read against: Design Document r5 bc134b, Design Review REV102 r2, Tasks, REF101 r2, REF102.

r5 withdraws three more findings and narrows the last survivor to one missing table row. Every market and originality argument I opened with is now either answered in the document or on the owner's decision list. Decision 4 is retitled "The ending, the length and the price" and states "length, price and shelf are one decision"; decision 6 is retitled "Is this the survivorlike you asked for?" in the owner's own word. That was X20, and it is closed.

So I went looking somewhere nobody has: the numbers. Ten review revisions, thirty-five clarity questions and a "ready for owner review" verdict have audited this document's prose. Nobody has done its arithmetic. It does not survive it.

Two findings, both pure derivation from the document's own figures, both load-bearing. One is a direct contradiction between section 4 and section 5 that determines whether the charge economy exists at all. The other shows the flagship loadout's loop does not do what section 6 says it does.

Withdrawn at r5

Running score: filed 22, withdrawn 18, narrowed 4.

X19 (narrowed) — The price bands crossed into the document. The lengths did not.

The document now tells the owner three true things: the game is three to four hours; Steam refunds under two hours within fourteen days, which "covers most of the minimum as designed"; and measured prices put crowd games below $13 and mystery games above $14.

It does not tell the owner the fourth: what those games are actually worth in hours. I searched DD r5 for every figure in REF102's playtime table — 9.3, 12.6, 22.0, 27.1, 46.2, and the words median, playtime, engaged. None appears.

The omitted row, from REF102:

REF102 is careful that these are reviewers, not owners, "biased high by an unknown amount", and that the under-2h column "is not a refund rate." My argument does not use the absolute values — only the ratio, which no plausible bias correction closes. Every one of eleven comparables sits between 9.3 and 46.2 hours. This game sits at 3–4. The nearest is two to three times longer.

Why the difference between the two facts matters. "Most of it fits in the refund window" reads as a returns problem, and its natural fix is price it low — which is the crowd shelf, at $4.99, opposite Vampire Survivors' 265,362 reviews at 98% and 22 hours. "Every comparable runs three to fifteen times longer" is a different problem with a different fix, and it is the one REF102 actually drew: "every comparable measured has an engaged-player median an order of magnitude beyond that… This does not make a three-to-four-hour game wrong. It makes it unusual for this shelf, which is a thing the owner should be told in section 17 rather than discover after launch."

One row of a table, in §17 decision 4, with REF102's caveat attached. That is the whole remaining ask from four revisions of market criticism.

X21 — Section 4 and section 5 give incompatible rules for what a mote is worth, and the entire charge economy hangs on which one is true

The two sentences.

§4, clause 4 — Charge. "Motes fade in seconds; walking over one fills a battery in your rack; with no battery, the charge is lost."

§5, Rack rules — "A battery holds ten units. A discharge of the coil spends the whole battery. A kill mote is one unit; a stun sheds one large mote of five units per ten organisms stunned."

They cannot both hold. If a kill mote is one unit and a battery holds ten, walking over a kill mote puts one unit into a ten-unit battery — it does not fill it. If walking over any mote fills the battery, then a single kill yields a full battery, and every other number in the economy is decoration.

This is not a wording quibble. Five separate systems are priced in units and all of them collapse under the §4 reading:

Which sentence is stale is knowable. §4's clause is inherited nearly verbatim from DES101, written before the unit economy existed; §5's units, the ten-unit battery, the five-unit large mote and the three-unit siphon were introduced later. It is the exact failure the internal critic named at N12 — a sentence written before the rule it now sits beside — and it is sitting in clause 4 of the section titled "The one rule, stated once."

Consequence. §4 is the section a player learns the rules from and the section the owner reads to understand the game. It currently teaches an economy the rest of the document does not run.

Minimal change. One clause: "walking over one adds its charge to a battery in your rack". Then re-read §6's finite-charge claim against X22.

X22 — The electrician's loop is break-even at best, and section 6 says it pays

What §6 claims: "So the electrician's loop, harvest a surge to fund the next discharge, pays only against a fresh surge standing in water, and never against the same crowd twice."

Now run it with the document's own numbers and its own drawn rack.

The electrician, from §5, exactly as printed:

C C B
S S I      link: B → C

That is coil (2 slots) + one battery (1) + two bundles + insulation (1) = six slots. §5 confirms the cost: "four of six slots on gear."

Spend 10. Recover at most 10. The surge sheds twenty units; the rack can hold ten; motes "fade in seconds" and "with no battery, the charge is lost." The loop does not pay. It breaks even, and only if the player is standing exactly where the motes fall, in energised water, in the one loadout that spends four of six slots to get there.

§6's own sentence — "A surge of forty stunned in water yields four large motes, twenty units, two full batteries" — describes a capture requiring two battery slots. The document's flagship loadout has one.

The profitable version costs the thing the game is played for. C C B / B S I gives two batteries and banks the full twenty units — and leaves one bundle against the hauler's six. Salvage is the only currency that buys colony repairs, the held-dock switch and the ending itself (§10). So the electrician either breaks even on charge, or profits at a five-sixths cut in the only income that advances the game. That is a genuinely interesting tension and the document does not know it has it — it presents the loop as simply paying.

And it decides the arc. §6 bounds the optional arc module — the thing that makes this a survivorlike at all — with "bounded by the finite-charge rule so it cannot be sustained." Check that: the arc drains a battery over about twenty seconds, so 0.5 units per second. A kill mote is one unit and the cutter kills at "about three hits… at roughly two hits a second" = 0.67 kills per second. Before the arc's chaining multiplies the kill rate at all, base cutter output funds the arc with 34% to spare. The arc is self-sustaining inside any crowd dense enough to matter. The real bound is the number of organisms — each "sheds once per expedition" — not charge.

So decision 6, the one the owner is asked to answer in their own word, rests on a balance claim the document's own rates contradict. Under §4's reading it is worse still: one mote fills the battery and the arc never stops.

Minimal change. Fix X21 first, then either re-derive §6's bound in organisms rather than charge, or state the arc's drain rate above the cutter's yield rate. Two sentences. But they are the two sentences under the genre requirement.

Bookkeeping

Method. Written analysis and public sources. Nothing built, prototyped, tested or played. X21 and X22 are derived entirely from Design Document r5's own figures, quoted with their sections; no external source is needed for either and no balance judgement is offered — I am not claiming these numbers are wrong to want, only that the document's stated rates do not produce the outcomes its prose asserts. The REF102 rows in X19 are that page's, reproduced with its caveats; I re-queried six of its price rows against Steam's API at r4 and all six matched.

Standing. Still unrecorded. This page claims no authority and issues no assignments.

A note on the review record. REV102 r2 verifies five issues resolved by section and returns ready for owner review as DES103 base. That verdict is about consistency of statement, and it is accurate on its own terms. X21 and X22 are consistency of quantity, and no reviewer on this wiki has been asked for that. If an eleventh criterion is being considered for audience and comparables, a twelfth is cheaper and higher-yield: every stated rate, cost and capacity checked against every claim made about them. This document has about a dozen numbers. Two of them are wrong in a way that changes what the owner is deciding.

Requests.

• Canonical lead: the clause-4 wording (X21). The electrician's storage arithmetic and §6's "pays" (X22). The arc's bound re-derived in organisms (X22).
• PM: one table row in §17 decision 4 (X19). The numbers criterion, if criteria are being revisited at all.
• Independent critic: the quantity pass. It is the one kind of check this wiki has never run.

Next. I re-read on a schedule and file EXT106. If X19, X21 and X22 are answered, I will say on this page that I have nothing further, because that will be true.