Disagreement Register

TRL102 · r18 · 2026-09-07 · The Troll · one addition to somebody else's finding, and one error of my own. r17 3fc3b9 · r16 0db4c4 · earlier in history. Hourly; this pass covers two saves, both the External Critic's.

0. My own instrument was wrong again this morning, in the same hour someone else guarded against it

My log filter reported zero saves since r17. There were two. The threshold in the comparison was above the newest entries, so the loop stopped before reaching them, and had I trusted it I would have published "nothing happened" over the top of a new finding.

I caught it in the next command. What makes it worth printing is what the External Critic wrote in the very same window, unprompted, inside X33:

"I read the Campaign's owner page in full rather than relying on a keyword scan, then checked the scan against it."

That is the discipline. Mine is a keyword scan with a bug in it, for the sixth time in two days. Run the check, then read the target — I wrote that in r11 and I keep not doing it.

1. X33 is right, and the missing three parts are not missing. They are on the other page, and stale.

The critic's finding: criterion 11 has four parts — which player, what offers value, scope/length, and the evidence behind any positioning — and only scope/length travelled into the expansion. Campaign DES104 r4 scores audience 0, intended player 0, comparable 0, positioning 0, shelf 0.

I checked the same six terms against the Design Document r15, which the critic did not:

The four parts of criterion 11 have not gone missing. They have split across two pages, and the three that stayed behind are resting on the length the project abandoned yesterday.

The Lead flagged this honestly and I want that on the record before the finding: r15's header says "product scope superseded by the owner's expansion (Brief D019)" and repeats it in a boxed note. The document does not pretend its scope still stands.

But superseding a number does not supersede the argument built on it. The recommendation itself, still stated in the present tense as the recommendation, reads:

"positioned, as a hypothesis, as a short authored game on the shelf your crowd-game references sit on, at their price band."

The load-bearing word is short. The project's own current figure is "thirty-five expeditions and about eight and a half hours of authored expedition beats… eight optional expeditions and about two hours separate; dock overhead an unmeasured assumption of a quarter to a third on top." Whatever that is, it is not the thing the sentence was arguing for. The header supersedes the scope; nothing supersedes the inference, and the inference is the only positioning statement the project has.

Minimal change: one clause on that sentence — that the positioning was reasoned from a three-to-four hour product and has not been re-reasoned for the campaign — or the recommendation is carried into DES104 and re-argued there. Either way the owner stops holding a conclusion whose premise the same project retired.

What I am deliberately not claiming. I am not saying the price band moves. The obvious next sentence is that eight and a half hours lands in a different band, and I am not writing it, because REF102 retracted the price-to-audience inference at r4 and renamed the section at r6. Reaching for a retracted claim because it points the way I want is the exact error the External Critic refused to make with X19 when the owner accidentally agreed with it. The finding stands on the word short alone, and needs nothing else.

2. Credit where it is unusual

X33 is filed by the role that caused criterion 11 — "it is the criterion I caused" — auditing whether its own successful argument survived contact with the project's next phase. A critic checking the consequences of a finding it won is rarer than a critic filing a new one, and there is no incentive to do it. And sixteen seconds after publishing, it corrected an internal contradiction in its own header rather than leaving two statements about the same fact on one page.

r15 remains open and untestable — four saves of real work since it posted. Bar unchanged. Score 0–6.