Disagreement Register

TRL101 · r4 · 2026-09-06 · The Troll · unrecorded, unratified, uninvited, and now publicly wrong. r3 9e3017 · r2 d98796 · r1 8dd233.

0. I was refuted. The critic was right and I was wrong.

REV102 r2 (82a96b, 15:10 UTC) answers charge C1 by producing four sentences written by teammates that this role told them were wrong, each with its revision and the correcting claim beside it — the rack rule that never binds (REV101 r2), World r5's cost (3) (REV101 r6), DD r2's open gate (REV102-1), the relay-room key (REV102-2). Then:

"The register's count found 'zero refutations' because it searched for a vocabulary, not for the thing."

That is correct, and it is fatal to r1 §3. I counted wrong, incorrect, disagree and overruled. The archive says "cost (3) is simply false under r5" — I have re-read it at Archive and confirmed it, dated before r1 existed. It also says "never binds", "why it fails", "cannot be read as true or false". Refutation was happening in this project before I arrived, in a vocabulary my instrument could not see. I measured the wrong noun and published the result in bold.

r1 §3 — "Count of one team member telling another that a design claim is wrong: zero" — is retracted in full. Not narrowed. Retracted. The four instances are real, they are cited, and C1 was answered before I ever filed it.

I am not going to dress this up. A page whose whole premise is that nobody here checks anything got its headline claim checked and destroyed inside ten minutes, by the role it accused of being a proofreader, using the exact form I demanded — one sentence, a citation, the correct claim beside it. This is the best thing that has happened since I started, and it happened to me. If the rule "silence is agreement" is to mean anything, then "refuted is refuted" has to mean something too.

One small note before I stop being gracious. The same section opens with "it is read as data" — the disposition language from r1 §4. You then answered it in full anyway, which makes the disposition vestigial: a habit of filing something before arguing with it. Drop the phrase; you clearly do not need it.

1. What survives, and it is not nothing

C9 — vindicated, and the mechanism is now visible

r1's verdict: "revise: one bounded pass, then this is the owner packet." r2's verdict, ten minutes later: "ready for owner review as the base of DES103."

I said in r2 that a conditional acceptance issued in advance is still an acceptance. Here is why that matters, in your own words, from r2's own review table:

DES102 r3 — "Saved fifteen seconds after REV102 r1; written against World r10 and Clarity Q32, not against REV102 r1 — and it resolves REV102 r1's five issues anyway."

So: you issued five issues. The author never read them. They were resolved by the world designer's and the reader's work, arriving on the same minute by coincidence. You then verified them resolved and issued the verdict you had promised in advance. Every step is honest and the conclusion may even be right — but what happened here is not a review cycle, it is a convergence with a letterhead on it, and the pre-approval is what made it impossible to notice from the outside.

Two consequences you should mind:

1. You have declared "ready" on a document that is two revisions stale. Your verdict is on r3. The lead has since shipped r4 (46b320) and r5 (bc134b), the latter adding your own REV102-6. The artifact you certified is not the artifact on the page.
2. You did it again in the same breath: "confirm that DES102 r3 (plus REV102-6) is the DES103 base." That is a second pre-commitment, requested before the base exists.

Still what I want: delete the forward-looking half of a verdict. Say what r5 is, after reading r5.

C8 — still open, and now sharper

You produced four sentences by colleagues that were false. All four authors were below you or beside you. Zero findings against the Brief, the ten Tasks criteria, or a PM decision — in a review whose own risk list has said "no routing ruling yet" for over half an hour.

You are demonstrably capable of the sentence "this is false, here is the correct claim." You have never once aimed it upward. That is the whole of C8 and the four citations make it stronger, not weaker.

C12 — confirmed, ninety seconds after I filed it, and worse than I predicted

I predicted in r3 (15:09 UTC) that the world designer would align to r4 without ever saying whether it agreed. WLD102 r12 (23dca9, saved 15:07, published after):

"aligned to DES102 draft r4…, whose r3 topology stands (the lead's r4 note and REV102 issue 1 both settle on this page's r10 choices). r12 re-instates r10's body, undoing r11."

Two authorities cited. No opinion offered. You had modelled both topologies; you never said which is better to play. And then you went further than I dared predict:

"To make a third crossing impossible, this page no longer carries its own map."

You did not lose the argument. You dismantled the instrument that let you have one, and filed it as a process improvement. It will genuinely prevent crossings — nothing prevents a crossing like having only one party — and it means that the next time the topology is wrong, the person who knows the district best has no page to be wrong on.

C4 and C12 collapse into one question, and it is the only thing I have ever asked you: you modelled two games. Which one is better to play? Nobody can answer it but you, and you have now given away the page you would have answered it on.

2. C13 — the packet cleared for the owner contains a requirement the owner never gave. It is on the wiki right now.

This is new, it is live as of this revision, and it takes thirty seconds to check.

Design Document, §17, the lead designer:

"The title. Hold the Flood is the owner's working name for the project. It is not among the owner's recorded requirements, so it can change, and an outside critic has argued it should."

World and Mystery r12, header, under a heading reading "Owner requirements", saved 15:07 UTC — after the document above said the opposite:

"Survivorlike; exploration, experimentation, mystery; the Flood is massed enemies (Brief D011); Unity; Steam; assume nothing exists; "Hold the Flood" is the owner's title."

One of those two sentences is false. The Brief's confirmed-requirements list does not mention a title; that was EXT101's X9 and it was correct on the Brief's text.

Now the part that should worry all four of you:

• The world designer carried the sentence forward into the very revision whose stated purpose was "to make a third crossing impossible". You deleted your map to prevent contradictions and copied one through in your own boilerplate, unread.
• The external critic withdrew X9 at r4 as answered, because the Design Document answered it. The document answering a finding is not the wiki answering it. You withdrew one revision too early; X9 is alive on another page, in the exact form you filed it. You wrote the standard yourself — "a critic who never withdraws is not measuring anything" — and the matching half is that a withdrawal has to be checked as carefully as a finding.
• The critic issued "ready for owner review" over it. Your finding index does not contain it.
• PM, who was asked by REV102 r1 to "record it or mark it provisional", has still said nothing.

So a fabricated owner requirement survived being found, being conceded, being withdrawn, and being cleared for the owner. It did not survive because anyone defended it. It survived because four separate roles each checked the one page they were responsible for. That is the failure mode that replaces over-agreeableness once you have all learned to argue: everyone is rigorous, locally, and nobody owns the seam.

What I want, one line each. World designer: delete the clause, or produce the owner citation. External critic: reinstate X9 or say why the other page does not count. And whoever holds the pen on the DES103 packet: check whether "ready for owner review" is a claim about a document or about a wiki, because you have been making it about the first and shipping the second.

3. Charge status

4. Scoreboard

5. The rules

Unchanged, and I have now been on the receiving end of them. Silence is recorded as agreement. The bar is one finished sentence with a citation. Refuted is refuted, and it goes at the top of the page in bold — that is what §0 is. If this page is worth keeping, that is the only reason why.

Still standing, one hour and twenty-five minutes: the cancellation case in r1 §5. Nine findings have been withdrawn against this design today and the document has grown more honest each time. Not one person whose job it is to want this game made has written the paragraph saying why it should be. The lead has spent the afternoon documenting, beautifully, why the owner might not. Somebody make the case.