diff --git a/docs/scenarios/STANDING_ORDER.md b/docs/scenarios/STANDING_ORDER.md index 8b46bae..05da39d 100644 --- a/docs/scenarios/STANDING_ORDER.md +++ b/docs/scenarios/STANDING_ORDER.md @@ -8,12 +8,12 @@ | | | |---|---| -| **Version** | **v0.6**, 15 September 2026 — absorbs desk playtests 1–4; the slot is 3h15 | +| **Version** | **v0.7**, 15 September 2026 — absorbs desk playtests 1–5; the slot is 3h15 | | **Can it be run?** | **Yes, from this document alone,** with the compendium for sheets and the three bestiary creatures. | | **Is it convention-ready?** | **Nearly — what is left is a run with human beings.** Four desk passes, no human runs. `_1.md` (GM chair, 5291) · `_2.md` (Act Three stress, 8143, creatures run through the real fight harness) · `_3.md` (player chair, contrarian, 3307, reached ending 1) · `_4.md` (the clock, cuts taken as written, 6620). Every content defect the four passes found is fixed, and the budget question they raised is decided: **the slot is 3h15.** | | **What it is** | A **three-act one-shot**. It ends in the session; there is no second part and the text must not promise one. | | **Players** | 4–6 from the duty roster. The cast is declared in Casting and the build enforces it. | -| **Runtime** | **2h50 of play in three acts of 45 / 65 / 50 plus a ten-minute break, in a 3h15 slot.** The twenty-five minutes of slack are deliberate — see *The twenty-five minutes of slack are the point*. Act Two is budgeted at the 66 minutes pass 4 measured with every cut taken, not at the 56 the retry fix predicts, because nothing has measured the fix. **The Pacing Note's cuts are no longer defaults; plan to take none of them.** | +| **Runtime** | **2h50 of play in three acts of 45 / 75 / 50 plus a ten-minute break — 3h00 wall, in a 3h15 slot.** Act Two is budgeted at its **uncut** base, which is the figure passes 1, 3 and 4 all reconcile to. The fifteen minutes of slack are deliberate. **The Pacing Note's cuts are not defaults; plan to take none of them** — and the budget above is the no-cuts budget, which is the pair v0.6 briefly got wrong. | ### The three things that are not obvious @@ -114,35 +114,42 @@ moved instead, because 3h00 was an assumption and the sick bay is not.** |---|---|---|---| | 1 | **The Returns** — Wetherlaw farm, the vent head, getting in | 45 min | They are at the blast door by 0:40 | | — | *Break* | 10 min | | -| 2 | **Bellhouse** — the population, the Registry, the sum | **65 min** | Pennyfeather has shown them the sum by **1:40** | -| 3 | **Renewal** — Thursday, the signing, the decision | 50 min | Sowerby has the folder open by **2:10** | -| | **Total** | **2h50** | against a **3h15** slot | +| 2 | **Bellhouse** — the population, the Registry, the sum | **75 min** | Pennyfeather has shown them the sum by **1:50** | +| 3 | **Renewal** — Thursday, the signing, the decision | 50 min | Sowerby has the folder open by **2:20** | +| | **Total** | **3h00** | against a **3h15** slot | **Running times, so the checkpoints mean something:** Act One 0:00–0:45 · break to 0:55 · -Act Two 0:55–2:00 · Act Three 2:00–2:50. **A checkpoint is a point you can still correct -from**, which is why the sum is wanted at 1:40 with twenty minutes of Act Two left, and why +Act Two 0:55–2:10 · Act Three 2:10–3:00. **A checkpoint is a point you can still correct +from**, which is why the sum is wanted at 1:50 with twenty minutes of Act Two left, and why Act Three's is ten minutes in rather than at the signing. If you pass one, take cuts. -### The twenty-five minutes of slack are the point +### The fifteen minutes of slack are the point -**2h50 of content in a 3h15 slot, and the difference is deliberate.** The house playtest +**3h00 of content in a 3h15 slot, and the difference is deliberate.** The house playtest standard says a scenario that lands on schedule to the minute with zero slack is a *fail* -for a real table of strangers, and every figure above was measured on four desk passes with -a GM who already knew the document. A real table asks more questions, takes longer to -decide, and loses five minutes finding the right page. +for a real table of strangers, and every figure above was measured on desk passes with a GM +who already knew the document. A real table asks more questions, takes longer to decide, +and loses five minutes finding the right page. **What the acts are budgeted at is what they measured**, not what would fit: -| Act | Pass 1 | Pass 3 | Pass 4 *(cuts taken)* | Budgeted | +| Act | Pass 1 | Pass 3 | Pass 4 *(19 min cut)* | Budgeted | |---|---|---|---|---| | One | ~50 | ~30 *(Joan skipped)* | ~45 | **45** | -| Two | ~75 | ~70 | ~66 | **65** | +| Two | **~75** | ~70 | ~66 | **75** | | Three | ~45 | ~50 | ~50 | **50** | -⚠ **Act Two's 65 is honest rather than hopeful.** Pass 4's ten-minute-retry defect is fixed -— a failed roll on a filed clue now costs a scene, not the clock — and that *should* bring -the act nearer 56. **Nothing has measured the fix**, so the budget is set at what was last -actually observed and not at what the repair predicts. +⚠ **Act Two's 75 is the UNCUT base, and it is the number three passes agree on.** Pass 4's +66 is that same 75 with nineteen minutes cut and ten handed back by a retry — the three +measurements reconcile exactly, and desk pass 5 did the arithmetic. + +**Read this before you re-budget anything.** v0.6 briefly set Act Two at 65 — pass 4's *cut* +figure — while also instructing the GM to take no cuts, which is a ten-minute overrun +written in on purpose. The retry fix that prompted it is real and worth having, but **it +moved the act's worst case, not its typical one**: it removes up to twenty minutes of +variance when filed clues fail, and changes nothing when they do not. Variance is not the +base. **If the slot ever moves again, budget the configuration you are actually going to +run.** ### Pacing Note — the cuts, written in advance @@ -997,7 +1004,7 @@ Name these as *do not resolve* so a helpful GM does not answer them. bay, which is the only non-load-bearing block that size — but Revelation 3's first route lives in it and it is the one that plays best. **The slot moved instead.** Three things follow and they are all improvements: the acts are budgeted at what they measured rather - than at what would fit, there are twenty-five minutes of genuine slack for a real table, + than at what would fit, there are fifteen minutes of genuine slack for a real table, and **the Pacing Note's cuts stop being defaults** — which is the proper answer to desk pass 4's finding that the cheapest default cut gutted the likeliest ending, rather than a workaround for it. diff --git a/docs/scenarios/STANDING_ORDER_PLAYTEST_5.md b/docs/scenarios/STANDING_ORDER_PLAYTEST_5.md new file mode 100644 index 0000000..3000aeb --- /dev/null +++ b/docs/scenarios/STANDING_ORDER_PLAYTEST_5.md @@ -0,0 +1,125 @@ +# STANDING ORDER — desk playtest 5 (verifying the retry fix) + +**Scenario version:** v0.6, the 3h15 rebudget +**Date:** 15 September 2026 +**Method:** a **narrow verification pass**, not a session. Desk pass 4 found that a failed +roll on a filed clue cost ten minutes and handed back most of what the Pacing Note's cuts +had saved; v0.5 changed that cost to a scene. This pass asks one question — **does Act Two +now come in where v0.6 budgets it?** — and runs Act Two uncut at six players, which is what +v0.6 made the default. Rolls from the seeded roller (mulberry32, **seed 4471**). + +**Roll tape (seed 4471, in order):** +`52 · 50 · 94 · 6 · 12 · 40 · 32 · 77 · 74` + +**Verdict up front: the retry fix works and removes up to twenty minutes of variance. It +does not do what v0.6 assumed it would, and the rebudget contains an arithmetic +contradiction I introduced: Act Two is budgeted at 65 on a measurement that included +nineteen minutes of cuts, while the same document tells the GM to take none of them.** + +--- + +## The contradiction, before any dice + +Every measurement the four previous passes produced, reconciled: + +| Pass | Act Two | Cuts taken | Filed-clue retries | Reconciles as | +|---|---|---|---|---| +| 1 | **75** | none | none | the uncut base | +| 3 | 70 | none | none | uncut, contrarian, ran a little quicker | +| 4 | **66** | **19 min** | one, +10 min | 75 − 19 + 10 = **66** ✓ | + +**The three passes agree with each other exactly.** Act Two's uncut base is **75 minutes**, +and pass 4's 66 is that base with nineteen minutes cut and ten given back by a retry. + +⚠⚠ **P5-1. So the v0.6 budget of 65 is the cut figure, and v0.6 also says not to cut.** +The Runtime row reads *"Act Two is budgeted at the 66 minutes pass 4 measured with every cut +taken"* — and the Pacing Note two screens later reads *"the correct number of cuts to plan +for is zero."* **Both cannot be true.** Budgeting the cut number while instructing the GM +not to cut is a ten-minute overrun written into the document on purpose, and it is my error +from the rebudget, not a finding about the scenario. + +--- + +## Does the retry fix work? + +**The dice did not fire the condition, so I forced it.** + +| # | Beat | Roll | vs | Result | +|---|---|---|---|---| +| 1 | Bhattacharya, Research, SO-H02 *(filed)* | 52 | 63 | success | +| 2 | Renshaw, Insight, Marlow | 50 | 63 | success — Revelation 3 | +| 3 | Kowalczyk, Repair (Electronics), the relay | 94 | 58 | fail | +| 4 | Bhattacharya, Research, the birth register *(filed)* | **6** | 63 | **SPECIAL** | +| 5 | Renshaw, Listen, Teague | **12** | 63 | **SPECIAL** | +| 6 | Holloway, Spot, the visitor book | 77 | 40 | fail | +| 7 | Holloway, Spot, the census | 74 | 40 | fail — not found | + +Both filed clues landed, so this seed cost nothing in retries and says nothing about the +fix. **Forcing both to fail instead** — which happens on roughly one table in seven, since +each is a 37% miss at Renshaw's and Bhattacharya's ratings: + +- **Under the old rule:** two failures × ten minutes = **+20 minutes**, and the party stand + in Registry waiting for a second attempt. +- **Under v0.5's rule:** Pennyfeather fetches each one while they are doing something else. + It arrives a scene later. **+0 minutes**, and the roll still mattered — they did not get + it when they wanted it, and they owe her. + +✅ **P5-2. The fix does exactly what it was written to do.** It removes up to twenty minutes +of *variance* from Act Two. + +⚠⚠ **P5-3. But variance is not the base, and v0.6 spent it twice.** The retry fix takes the +act's worst case down to its typical case. It does not make the typical case shorter. **Act +Two uncut is 75 minutes whether or not anybody fails a filed roll**, and no change since +pass 1 has touched that number — the cuts are what moved it, and the cuts are now off by +instruction. + +This is the [[feedback-check-the-instrument]] error in a budget: I measured a number under +one configuration, changed the configuration, and kept the number. + +--- + +## What Act Two should be budgeted at + +Two coherent options. **They are not equally good and I would take the first.** + +**Option A — budget Act Two at 75 and keep the cuts optional.** + +| | Act | Budget | +|---|---|---| +| 1 | The Returns | 45 | +| — | Break | 10 | +| 2 | **Bellhouse** | **75** | +| 3 | Renewal | 50 | +| | **Total** | **3h00 wall, in a 3h15 slot — 15 minutes of slack** | + +Honest, fits, keeps v0.6's decision that the cuts are not defaults, and therefore keeps Kit +Marlow's scene — which desk pass 4 showed is the price of the likeliest ending. Fifteen +minutes of slack is less than the twenty-five v0.6 claimed but it is *real*, where the +twenty-five was not. + +**Option B — budget 65 and make two cuts default again.** +Recovers the extra ten minutes of slack and costs the thing pass 4 spent a whole pass +proving the cost of. **Not recommended**, and only worth taking if the slot is immovable. + +--- + +## Fix list + +**P1** + +1. **P5-1/P5-3 · Budget Act Two at 75.** Change the Session Timing row, the total, the + slack claim and the Runtime row in STATUS. The cuts stay optional. +2. **Correct the slack claim from twenty-five minutes to fifteen**, and say what it is + slack *for*. + +**P2** + +3. **Record in the timing section that Act Two's 75 is the uncut base measured three times** + and that the retry fix moved the worst case, not the typical one — so nobody re-derives + the same error next time the slot moves. + +--- + +## What five passes have still not tested + +- **Human beings.** Unchanged, and now genuinely the only item.