RingBRP/docs/scenarios/CLEAN_GROUND_PLAYTEST_8.md

CLEAN GROUND — desk playtest 8 (the cold read, and the outcome nobody wrote)

Scenario version: v0.18, at commit 8a051f7 Date: 13 September 2026 Method: two halves. A cold read — the document taken as a GM meeting it thirty minutes before a slot, measured rather than impressioned: section sizes, counted claims, cross-references, and whether the STATUS block still tells the truth. And an honest session, six players, seeded roller, seed 5108, nothing forced. Earlier passes: 4271, 9314, 5580/7731, 6142, 8815, 3319, 7742.

Why this pass. Three versions of fixes (v0.16, v0.17, v0.18) have landed since anything was played, and every one of them was written by somebody who already knew the document. The only thing left before a human run is whether a GM who does not know it can run it from the page — which no dice pass can reach and which a human run will punish first.

Verdict: the most likely outcome of the scenario's climactic roll is not written down. It is not the one that is taken and it is not the one that is unsettled; it is the agent simply winning, which happens 48.1% of the time against the default target, and the document describes the other two and not that one.

---

The rolls, in order

seed 5108 — no forcing

ACT ONE
  Bhattacharya  Research — Corrigan's paperwork               92 vs  63  FAILURE
  Braithwaite   Medicine — the body                           87 vs  53  FAILURE
  Renshaw       Persuade — Swinburne                           8 vs  63  SPECIAL
  Pollard       Spot — the stump                              39 vs  53  success
  Bhattacharya  Knowledge — the squared notch                 53 vs  63  success

ACT TWO
  Renshaw       Persuade — the column                         52 vs  63  success
  Pollard       Spot — the same six at the back               86 vs  53  FAILURE
  Ashcroft      Insight — it is offered, not remembered       19 vs  63  success
                → ASHCROFT REFUSES THE OFFER (third pass running)

ACT THREE
  Bhattacharya  Anomaly Lore — the sky has not moved          20 vs  63  success
  Bhattacharya  Research — RESERVE ENTRY (NORTHERN)           49 vs  63  success
  Ashcroft      Anomaly Lore — what a peg is                  30 vs  63  success
  hollow man swap  75 rolled 56 success  vs  Okonkwo 55 rolled 17 success
                   → resisting. OKONKWO HOLDS. Third pass running.

ACT FOUR
  Braithwaite   Xenology (Baseline) — the understudy          28 vs  53  success
  Renshaw       Insight — it is word-perfect                  38 vs  63  success
  Renshaw       Psychology — it thinks it is the file          5 vs  40  SPECIAL
  COUNTDOWN STEP 5
  understudy 75 rolled 23 success  vs  Okonkwo 55 rolled 19 success
                   → RESISTING WINS OUTRIGHT. Nobody is taken.

The luckiest seed in eight passes, and it walked straight into the gap. Nobody is taken, which is neither of the two outcomes Act Four writes.

---

⚠ FINDING 1 — the climactic roll has three outcomes and the document describes two

Step 5 of the countdown is the moment the case has been building to. opposedContestFor returns three things, and only two of them are on the page.

Enumerated over all 10,000 roll pairs — no seed, reproducible from the rule:

Step 5 against	Taken	Resisted outright	Unsettled
Okonkwo, POW×5 55 — the default target	40.7%	48.1%	11.3%
Braithwaite, 60	37.6%	52.4%	10.0%
Ashcroft, 85, refused	21.9%	74.3%	3.8%
Ashcroft, 85 → 42, accepted	48.7%	36.8%	14.5%

The taken column and the unsettled column are both printed in the document and both reproduce exactly, so the source is right and has always been right. The middle column appears nowhere. Against the default target it is the single most likely result of the scenario's climax, and against an Ashcroft who refused — which is what most tables will have, as v0.17 now says — it is three outcomes in four.

• ⚠ And the framing makes it worse than an omission. The paragraph under the table is headed "WHEN STEP 5 COMES TO NOTHING" and is entirely about unsettled: "the rule returns unsettled, which is not the agent winning". It then gives the one piece of read-aloud the beat has — for a moment it is wearing the wrong face. A GM whose agent simply won has come to nothing too, finds the only nearby text, and reads a passage written for a different outcome. The document warns against confusing the two and then supplies nothing for the commoner one.
• This is not a rules problem. The rule is right, the tie goes to the resisting side, and GM ESSENTIALS says so. It is that the scenario priced the reach in how often it succeeds and never wrote the far likelier case where a person holds.
• Fix direction, and it wants to be good rather than merely present. A clean resist is the only moment in the case where somebody beats the thing on their own, and it should not be a shrug. The material is already there: the understudy has practised for thirty years and it is not stupid — it takes the easiest file. Being the file it could not open is the one win the party gets all night.

---

⚠ FINDING 2 — EXPOSURE's five things are six, and they run 1, 2, 3, 4, 6, 5

Introduced in v0.16 (aa3ae24) — my own commit — and survived two versions, five guard additions and a peer's sweep without anybody noticing.

   999  #### The five things a GM needs from that table
  1001  1. Do not fix the understudy
  1006  2. Adding hollow men buys rounds, not danger
  1013  3. The column is the lethal encounter
  1020  4. The four-player cut is a different game
  1070  6. A fight costs the session          <- inserted here
  1096  5. Their sheets read twice as effective as they are

The insertion anchored on the end of item 4's block, which in the file sits before item 5 rather than after it. The edit asserted a unique match and got one; a unique match protects against editing the wrong text and not at all against inserting in the wrong place.

• No guard can see this and none should be built for it. It is a numbered list in prose. What it argues for is the cold read itself: seven passes of dice found nothing here, because dice never read a heading.
• The heading also still says five.

---

FINDING 3 — EXPOSURE is 18% of the document and larger than any act

Measured, not felt:

Section	Words
EXPOSURE	3,176	18% of the document
Act Two	1,635	the longest act
Act Four	1,267	the act EXPOSURE sits inside
Act One	1,127
Act Three	1,050
GM ESSENTIALS	605
Whole document	17,776	~90 minutes to read once, carefully

EXPOSURE is two and a half times the act it is filed in, in a scenario whose stated thesis is that the fight is not the point. Its own heading concedes the filing is wrong — "Read before Act Two, not before Act Four."

• This is a judgement, not a defect, and it is recorded as one. Every figure in there was added because a pass found a GM would get something wrong without it, and I added the largest single block of it in v0.16. Nothing in it is padding.
• But 3,176 words is sixteen minutes, and it is sixteen minutes about an encounter the document spends its whole length telling the GM not to have. A convention GM with thirty minutes of prep who reads EXPOSURE has read half their prep budget on the fight they are being told to avoid.
• Fix direction: not deletion. A hundred-word summary at the top of EXPOSURE that a GM can read instead, with the body left as reference for the GM who has time or whose table has already reached for a gun. The five — six — numbered lessons are nearly that already; they are just at the bottom, after the tables.

---

FINDING 4 — an open question the document has already answered

"Scent is unusable... I moved Act Four's working test to Xenology (Baseline), which puts Braithwaite on the critical path. Worth deciding whether you want the whole scenario's one reliable test sitting on a single sheet."

It no longer is. v0.14 made the tell un-gated — "Anyone who examines him properly finds this, with no roll" — and desk pass 6 ran the act with both Xenology and Psychology failed and it held. The question was answered by the fixes and never closed, so it is still sitting in Open Questions asking the author to decide something that has been decided.

• Small, but it is the section where a reader is told what still needs their attention, and a stale item there costs more than its word count.
• The other open item — how hard do you want the PC-replacement, default NPC-first — is genuinely open and should stay.

---

FINDING 5 — the guard's 85% landed on the first honest pass after it was built

Renshaw rolled 5 against Psychology 40 — a special — on a beat that states a failure case and nothing above it. check-outcomes measures the chance of a session hitting a special on a beat with nothing written at 85%, and the very next session hit one.

• Not evidence of a rate; one session is one session, and the 85% is enumerated from the rule rather than counted from play. It is a demonstration that the number describes something real.
• The obvious special here is worth writing: it is not pretending to be human, it thinks it is the file is already the best line in Act Four, and a special should buy which file it thinks it is — that it corrects the agents on Prichard's service history, in the tone of a man who has read his own record more recently than they have.

---

What worked

v0.17's Swinburne special fired, and it is right. Renshaw rolled an 8. She produced the note, walked them to the stump, and talked on the way — the dog in June, the vet who wrote suspected poisoning. That is four months of the case's history arriving in Act One because somebody rolled well, and it is exactly the shape GM ESSENTIALS now prescribes: portable, private or early. The fix pass 7 asked for works and needs nothing.

The cold read found no broken cross-references. All six pointers (GM ESSENTIALS has, EXPOSURE has, See Act Two) resolve to sections that exist. Every counted claim except EXPOSURE's checks out: three things, three clues, four beats, six revelations, five acts — all correct.

Third pass running the swap has failed (75 against POW×5 55, and the tie goes to the resister), and third running Ashcroft has refused. Both are what the numbers predict — refusing needs a 63% Insight — and neither is a defect. Recorded so the next pass does not read a streak into them.

---

Timing

Not re-measured. Passes 4, 5 and 7 own the clock, and this pass forced nothing and cut nothing that would move it. Act One ran long on improvisation again — two failures and an 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 — 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.
3. Put a hundred-word summary at the top of EXPOSURE. 3,176 words is sixteen minutes about the encounter the document exists to prevent. Do not cut the body; give the GM in a hurry somewhere to stop.
4. Close the stale open question about Xenology. v0.14 un-gated the tell and pass 6 proved it; Braithwaite has not been on the critical path for four versions.
5. Write the special on Act Four's Psychology — which file it thinks it is.

Still untested after eight passes: a fight the players choose; Ashcroft accepted and 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.