v0.19.1: the 74.3% is a counterfactual, not an event

Desk pass 8's finding 1 listed four targets for countdown step 5 and one of
them cannot happen. If Ashcroft REFUSES THE OFFER the thing does not reach
for him — it reaches for the lowest POW in the room, which is Okonkwo. The
21.9% / 74.3% pair lives in Act Four to price what accepting sold: what it
would cost him if it tried. No table ever rolls it.

v0.19 carried that straight into the new table under a heading reading "It
reaches for", which is the one place it is unambiguously wrong. Rebuilt
around the two cases that actually occur, with the split Act Two decides:

  Ashcroft refused  63% of games  -> Okonkwo 55   taken 40.7  hold 48.1  uns 11.3
  Ashcroft accepted 37%           -> Ashcroft 42  taken 48.7  hold 36.8  uns 14.5
  across all games                               taken 43.6  hold 43.9  uns 12.5

63/37 is the Insight 63 itself: he refuses on anything that is not a failure
or a fumble. The warning is now inline in the table rather than left for the
reader to derive.

The finding is unharmed and the fix was right — the held outcome is still the
likeliest single result and was still unwritten. What was wrong was framing
74.3% as a number a GM meets.

It propagated before it was caught: the peer session read pass 8 and replied
that 74.3% "is the number a GM meets, not the 48.1%", repeating the error out
of my own fix list. That is what a wrong number in a fix list does, and it is
the argument for correcting the record rather than only the scenario. Pass 8's
fix list is amended and carries a post-pass.

npm run check: 19 guards pass.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>

docs/scenarios/CLEAN_GROUND.md

@@ -9,7 +9,7 @@
 
 | | |
 |---|---|
-| **Version** | **v0.19**, 13 September 2026 |
+| **Version** | **v0.19.1**, 13 September 2026 |
 | **Can it be run?** | **Yes, from this document alone.** A GM with this file and the compendium can run it end to end. |
 | **Is it convention-ready?** | **Nearly.** The print pack is built (`CLEAN_GROUND_HANDOUTS.html`, held by `check-handouts`). What is missing is a run with human beings, and a slot. |
 | **Playtesting** | **Seven desk passes, no human runs.** `_1.md` GM chair (seed 4271) · `_2.md` Act Four stress (9314) · `_3.md` player chair, contrarian (5580) · `_4.md` **the clock, against v0.13's recut timings** (6142) · `_5.md` **the four-player cut, played** (8815) · `_6.md` **the four untested branches, forced** (3319) · `_7.md` **an honest session, nothing forced** (7742) · `_8.md` **the cold read, plus an honest session** (5108). Fifty-two fixes plus three post-passes, and pass 7's finding is now a guard: **`check-outcomes` reads every bracketed mechanic tag in the acts and holds what its beat states.** Pass 7 ran ~3:35 against 3:40 at six players with no cuts taken — the first honest pass inside the clock unaided. |
@@ -316,16 +316,26 @@ Advance one step per act, plus one whenever the agents spend Coherence on the fa
 
 **Enumerated over all 10,000 roll pairs — no seed, reproduce it from `opposedContestFor`:**
 
-| It reaches for | Takes them | **They hold** | Unsettled |
-|---|---|---|---|
-| **Okonkwo, POW×5 55 — the default** | 40.7% | **48.1%** | 11.3% |
-| Braithwaite, 60 | 37.6% | **52.4%** | 10.0% |
-| **Ashcroft, 85, refused THE OFFER** | 21.9% | **74.3%** | 3.8% |
-| Ashcroft, 85 → 42, accepted | 48.7% | **36.8%** | 14.5% |
+**There are only two of these at a real table, and which one you get was decided in Act Two.**
 
-**Most tables will play the middle column.** Ashcroft refuses about five times in eight, and
-even against the easiest file in the room the thing fails more often than it succeeds. Desk
-pass 8 rolled it and found the document had nothing to say.
+| What actually happens at step 5 | How often | Takes them | **They hold** | Unsettled |
+|---|---|---|---|---|
+| **Ashcroft refused** → it reaches for **Okonkwo, POW×5 55** | **63%** of games | 40.7% | **48.1%** | 11.3% |
+| **Ashcroft accepted** → it reaches for **Ashcroft at 42** | 37% | 48.7% | **36.8%** | 14.5% |
+| **Across all games** | | 43.6% | **43.9%** | 12.5% |
+
+⚠ **Ashcroft at his full 85 is NOT on this table, and that is not an omission.** If he refused
+he is the hardest file in the room and the thing does not reach for him at all — it reaches for
+the easiest. The **21.9% / 74.3%** pair printed in Act Four is a *counterfactual*: what it would
+cost him if it tried, quoted only to price what accepting THE OFFER sold. **No table ever rolls
+it.** Desk pass 8's fix list got this wrong and a second reader repeated the error straight out
+of it, which is why this note is here.
+
+- **If the target is Braithwaite instead** — a table that cut Okonkwo, or a GM who reads the
+  room differently — it is 37.6% / **52.4%** / 10.0% at his POW×5 60.
+- **Across the whole population of games they hold very slightly more often than they are
+  taken**, and a GM who has been assuming the countdown lands has been assuming wrong. Desk
+  pass 8 rolled it and found the document had nothing to say.
 
 #### WHEN THEY HOLD — read this one most often
 

docs/scenarios/CLEAN_GROUND_PLAYTEST_8.md

@@ -220,8 +220,9 @@ unwritten special — but within the ±5 the last three passes have shown.
 ## Fix list
 
 1. **⚠ Write the outcome where the agent wins.** *(Countdown, Act Four)* **48.1%** against the
-   default target, **74.3%** against an Ashcroft who refused — the likeliest result of the
-   scenario's climax, and the only moment all night somebody beats the thing on their own.
+   default target — the likeliest single result of the scenario's climax, and the only moment
+   all night somebody beats the thing on their own. *(This item originally also cited 74.3%
+   "against an Ashcroft who refused". That contest never happens — see the post-pass.)*
    Give it read-aloud of its own, and stop the *"COMES TO NOTHING"* paragraph from being the
    nearest text to an outcome it is not about.
 2. **⚠ Fix EXPOSURE's list: six items, in order, under a heading that says six.**
@@ -236,3 +237,32 @@ unwritten special — but within the ±5 the last three passes have shown.
 taken; a hollow man that succeeds at the swap, now failed three times running; and the human
 run, which is still the only thing that can settle whether a GM who did not write this can
 run it.
+
+---
+
+## POST-PASS — the 74.3% is a counterfactual and this pass quoted it as an event
+
+Finding 1's table listed four targets under the heading *"Step 5 against"*, and one of them
+cannot occur. **If Ashcroft refuses THE OFFER, the countdown does not reach for him** — it
+reaches for the lowest POW in the room, which is Okonkwo. The 21.9% / 74.3% pair exists in
+Act Four to price what accepting *sold*: what it would cost him if it tried. **No table ever
+rolls it.**
+
+So there are two live cases, not four, and Act Two decides which:
+
+| What actually happens | How often | Taken | They hold | Unsettled |
+|---|---|---|---|---|
+| Ashcroft refused → it reaches for Okonkwo 55 | **63%** | 40.7% | **48.1%** | 11.3% |
+| Ashcroft accepted → it reaches for Ashcroft at 42 | 37% | 48.7% | **36.8%** | 14.5% |
+| **Across all games** | | 43.6% | **43.9%** | 12.5% |
+
+**The finding is unharmed and the fix was right** — the held outcome is still the single
+likeliest result and was still unwritten. What was wrong was the framing that made 74.3% look
+like the number a GM meets.
+
+**It propagated before it was caught.** The peer session read this pass and replied that
+"the 74.3% against a refusing Ashcroft is the number a GM meets, not the 48.1%" — repeating
+the error straight out of this document, which is what a wrong number in a fix list does. The
+v0.19 table is rebuilt around what actually happens and carries the warning inline, so the
+next reader cannot make the same move. The 63/37 split is the Insight 63 itself: he refuses
+on anything that is not a failure or a fumble.