diff --git a/WORK-RECORDS.md b/WORK-RECORDS.md
index f501be7..0d293e6 100644
--- a/WORK-RECORDS.md
+++ b/WORK-RECORDS.md
@@ -19,6 +19,7 @@
 | task | GROUND-WP-0002-T03 | done | — | workplans/GROUND-WP-0002-rulings-clay-borg-is-blocked-on.md |
 | task | GROUND-WP-0002-T04 | done | — | workplans/GROUND-WP-0002-rulings-clay-borg-is-blocked-on.md |
 | task | GROUND-WP-0002-T05 | done | — | workplans/GROUND-WP-0002-rulings-clay-borg-is-blocked-on.md |
+| task | GROUND-WP-0002-T06 | todo | — | workplans/GROUND-WP-0002-rulings-clay-borg-is-blocked-on.md |
 | task | GROUND-WP-0003-T01 | todo | — | workplans/GROUND-WP-0003-playtest-semi-and-coalitions.md |
 | task | GROUND-WP-0003-T02 | todo | — | workplans/GROUND-WP-0003-playtest-semi-and-coalitions.md |
 | task | GROUND-WP-0003-T03 | todo | — | workplans/GROUND-WP-0003-playtest-semi-and-coalitions.md |
diff --git a/workplans/GROUND-WP-0002-rulings-clay-borg-is-blocked-on.md b/workplans/GROUND-WP-0002-rulings-clay-borg-is-blocked-on.md
index 214a640..84a71da 100644
--- a/workplans/GROUND-WP-0002-rulings-clay-borg-is-blocked-on.md
+++ b/workplans/GROUND-WP-0002-rulings-clay-borg-is-blocked-on.md
@@ -229,60 +229,6 @@ T01 may dissolve it.
 | `gr-e02-shared-ground` | **Confirm** — encode `Modes.csv` MODE_COOP as printed |
 | `gr-e04-coalitions` | **Confirm** — encode `Modes.csv` MODE_COALITION as printed |
 
-## Task: GR-E01 vs GR-S01 — the group cannot win below 5 seats
-
-```task
-id: GROUND-WP-0002-T06
-status: todo
-priority: high
-```
-
-**This is the sharpest finding the engine has produced, it is confirmed by
-play, and it has a reproduction.**
-
-The maintainer played several 3-player games on 2026-08-03 and could not
-win any of them. The engine says why, computed from its own constants:
-
-| seats | Problems dealt (GR-S01) | maximum achievable | GR-E01 threshold | |
-|---:|---:|---:|---:|---|
-| 2 | 2 | 3 | 5 | **unreachable** |
-| 3 | 3 | 6 | 7 | **unreachable** |
-| 4 | 3 | 6 | 7 | **unreachable** |
-| 5 | 4 | 10 | 9 | reachable |
-| 6 | 4 | 10 | 9 | reachable |
-
-**Claim every Problem in play, concede nothing, play perfectly — and at 2,
-3 and 4 seats the group still loses.** In SHARED GROUND (GR-E02), where
-the group total *is* the outcome, those games cannot be won.
-
-**It is not a dataset problem.** The stand-in the engine used gives
-3 / 6 / 10; this repository's `Problems.csv` gives 4 / 6 / 9. Both fall
-short of the same 5 / 7 / 9. No edition data reconciles them.
-
-**Reproduction:** `games/ground/src/lib.rs` →
-`gd0001_group_success_is_unreachable_below_five_seats`, run with
-`cargo test -p games-ground --all-features gd0001 -- --nocapture`. It
-reads `problem_priorities` (GR-S01) and `threshold` (GR-E01) out of the
-engine, so it cannot drift from the rules it is testing.
-
-### The ruling needed
-
-Two numbers disagree and **only one of them can be right**:
-
-- **GR-S01's deal** — 2 / 3 / 4 Problems by player count, from a scenario
-  of five. Should more of the five be in play at low seat counts?
-- **GR-E01's thresholds** — 5 / 7 / 9. Were these set for a game that
-  deals more Problems than GR-S01 deals?
-
-A third possibility worth ruling out explicitly: **the thresholds may be
-intended to be unreachable at low seat counts**, with SHARED GROUND being
-a 5–6 player mode and the smaller counts played in another mode. If so,
-say it, because nothing currently does and the engine will keep offering
-a game that cannot be won.
-
-**Until this is ruled the engine cannot claim a working scoring mode**, and
-`gr-e01-threshold-unreachable-2p` stays as a failing-in-fact record.
-
 ## Task: GR-E03 and GR-E04 have never been played to the end
 
 ```task
diff --git a/workplans/GROUND-WP-0004-the-threshold-ruling-rests-on-a-bad-premise.md b/workplans/GROUND-WP-0004-the-threshold-ruling-rests-on-a-bad-premise.md
new file mode 100644
index 0000000..5b3e9aa
--- /dev/null
+++ b/workplans/GROUND-WP-0004-the-threshold-ruling-rests-on-a-bad-premise.md
@@ -0,0 +1,101 @@
+---
+id: GROUND-WP-0004
+type: workplan
+title: "Reopen GR-E01: the void ruling rests on a premise clay-borg has corrected"
+domain: consumer
+repo: ground-game
+status: ready
+owner: bernd
+topic_slug: whynot
+created: "2026-08-03"
+updated: "2026-08-03"
+origin: correction
+origin_ref: GROUND-WP-0002
+---
+
+# The GR-E01 ruling was made on a false premise, and clay-borg supplied it
+
+GROUND-WP-0002 T03 ruled `gr-e01-threshold-unreachable-2p`:
+
+> **Void / overturn as a rules gap** — stand-in artifact. With
+> authoritative `Problems.csv` (**12 pts available**) the printed
+> thresholds (2p→5, 3–4p→7, 5–6p→9) stand.
+
+**The 12 is wrong, and clay-borg is where it came from.** Twelve points
+exist *in the file*; **GR-S01 deals 2 / 3 / 4 of the five Problems**, so
+the points ever in play are 4 / 6 / 9. clay-borg asserted the 12 in its
+CB-WP-0021 declaration, corrected itself one commit later after actually
+running the arithmetic, and by then the ruling had been made.
+
+**The gap is not a stand-in artifact. It survives the authoritative
+dataset**, and it has now been confirmed at the table.
+
+## The evidence, in three independent forms
+
+**1. Play.** The maintainer played several 3-player games on 2026-08-03
+and could not win any of them.
+
+**2. The engine's own constants.**
+
+| seats | dealt (GR-S01) | max achievable | GR-E01 threshold | |
+|---:|---:|---:|---:|---|
+| 2 | 2 | 3 | 5 | **unreachable** |
+| 3 | 3 | 6 | 7 | **unreachable** |
+| 4 | 3 | 6 | 7 | **unreachable** |
+| 5 | 4 | 10 | 9 | reachable |
+| 6 | 4 | 10 | 9 | reachable |
+
+Claim every Problem in play, concede nothing, play perfectly — at 2, 3 and
+4 seats the group still loses. In SHARED GROUND, where the group total
+*is* the outcome, those games cannot be won.
+
+**3. Both datasets agree.** The stand-in gives 3 / 6 / 10; `Problems.csv`
+gives 4 / 6 / 9. Both fall short of 5 / 7 / 9. **No edition data
+reconciles them**, which is the part the void ruling turned on.
+
+**Reproduction:** `gd0001_group_success_is_unreachable_below_five_seats`
+in clay-borg —
+`cargo test -p games-ground --all-features gd0001 -- --nocapture`. It
+reads `problem_priorities` and `threshold` out of the engine, so it cannot
+drift from the rules it tests.
+
+## Task: rule again, on the corrected numbers
+
+```task
+id: GROUND-WP-0004-T01
+status: todo
+priority: high
+```
+
+Two numbers disagree and only one can be right:
+
+- **GR-S01's deal** — 2 / 3 / 4 of five Problems. Should more be in play
+  at low seat counts?
+- **GR-E01's thresholds** — 5 / 7 / 9. Were they set for a deal larger
+  than GR-S01 makes?
+
+A third possibility to rule out **explicitly**, because nothing currently
+does: the thresholds may be *intended* to be unreachable at low seat
+counts, with SHARED GROUND being a 5–6 player mode. If that is the
+design, say so — the engine will otherwise keep offering a game that
+cannot be won, and a player will keep concluding they played badly.
+
+`gr-e01-threshold-unreachable-2p` stays as a failing-in-fact record until
+this is answered. It is **not** retired on the previous ruling.
+
+## Task: how a clay-borg premise reached a ruling unchecked
+
+```task
+id: GROUND-WP-0004-T02
+status: todo
+priority: medium
+```
+
+Worth one paragraph, not a project: the ruling was sound reasoning on a
+supplied number, and the number was wrong. The engine now attaches a
+**runnable reproduction** to findings of this kind (clay-borg CB-WP-0022),
+so a future ruling can be checked against an artifact rather than a
+sentence.
+
+State whether that is sufficient from this side, or whether rulings should
+cite the reproduction they relied on.