clay-borg/workplans/CB-WP-0045-the-table-names-what-it-shows.md

id	kind	title	status	state_hub_workstream_id
CB-WP-0045	product	The table names what it shows	done	4191bea7-f1ea-40b9-a957-600dabb298e2

Purpose

structural tier  S   (rendering only -- no rule, no dependency, no budget,
                      no artifact contract moved)
declared tier    S

First declaration after ADR-0021 retired the chaos roll.

The reports

Three sessions, make trials (notes now bind per game — ADR-0019 holds: g1+0, g2+35, g3+44):

"Ok, that seemed more interesting and we lost." / "We did it this time." — 2026-08-08 1818

The H2 dynamics land. Two things do not:

"The GROUND and DARVO logic is far from self explanatory. That needs to be improved." — 2026-08-08 1818

"As the cards are now located differently on the table i cant tell which is problem 1, which 2 and so on."

The numbering defect [CB-WP-0043] caused

Every move label says "Problem 7". The number was rendered on the card only as a fallback for a missing title — so it appeared exactly when it carried least information, and was absent whenever the card was identifiable enough to be worth pointing at.

That was survivable while Problems sat in one ordered row: position was the number. CB-WP-0043 scattered them by scope, deleted the row, and took the implicit index with it — without anything going red, because no test named the number.

A rendering change removed a fact the page had been carrying positionally. Not the ADR-0018 family (nothing computes the wrong subject); the nearer relative is CB-WP-0024 — a fact held by presentation alone does not survive the presentation changing.

The legibility defect: vendored and never shown

DARVO.csv carries mandatory_effect per stage. Actions.csv carries rules_text for the three GROUND modes. Both were vendored, both were used only as tripwires, and neither ever reached the player. The seat panel said DARVO Reverse; the move buttons offered GR / OU / ND.

F18's shape once more: the edition's own words were present in the repo and could not fire. The player was asked to choose between three modes at the point where the action card has already left the screen.

Task: put the number and the words on the table

id: CB-WP-0045-T01
status: done
priority: high
state_hub_task_id: "bba3d8cb-7aa2-4bee-b97f-6ff8681dbec2"

Controls:

• every Problem on the table shows its number — asserted over view.problems.keys(), so a card that renders without one goes red whatever the fixture holds;
• the explanation is the edition's sentence, not our paraphrase — the DARVO test re-reads DARVO.csv and asserts a prefix of mandatory_effect appears verbatim (ADR-0015);
• the GROUND modes explain themselves where they are chosen — present when ChooseGroundMode is legal, and absent when it is not. An always-on wall of rules is how a player stops reading them, so the negative half is asserted too;
• both explanations are <details>, so the account can be closed.

Done 2026-08-08. 78 tests pass; make all green; verified against a live make ground VARIANT=h2 page — badges 1 2 3 4 … and the header reading rules h2-scoped-problem-stress.

Not done here

• Nothing explains what a scope does. H2 is the variation under study, the marker now says whose Problem it is and where it sits — and no text on the page says that a scoped Problem presses only the owner's bond network. The player can see the placement and still not know the rule it encodes. This is the largest remaining gap and it is a ground-game question first: the scope rule is ours (Variant), not the edition's, so there is no vendored sentence to render.
• The trial log header does not record the variant. A note from an H2 session and a note from a baseline session are indistinguishable in the log, which is a real problem now that both are being played.
• The <details> are closed-by-default for DARVO, open for GROUND. Chosen, not measured; whether a player opens the closed one is untested.