v0.8.0.7 — the Salvage Yard had nothing to say, and phases too little time to read

The Salvage Yard was face up all along; its tile just read "a card". apply.ts
pushes a synthetic train-<n> id on trainScheduled, nothing in s.cards matches it,
and cardName fell through to its default — and since a train is scheduled several
times a Day that id is on top most of the time. Measured before touching anything:
the tile read "a card" from the opening frame through 60 pushes while its depth
climbed from 2 to 8. cardName resolves it now, in sim/view.ts, because this is a
name.

The engine half is filed as Gitea#23 rather than fixed here. reshuffleIfDepleted
sweeps the Salvage Yard back into the draw deck, so that synthetic id can be
shuffled in and drawn into a hand as an id with no card behind it. Eight games
driven to 4000 moves across eight seeds produced zero reshuffles, so it is latent;
there are two defensible fixes and the choice turns on what the synthetic id is
for, which is not a call to make while fixing a label.

And phases scale with the speed control again, damped to a third of the rate. They
were pinned in v0.8.0.3 because scaling them walled off a player's own turn; pinned
turns out to be too short to read at 10x. Damped satisfies both: 1x unchanged, 10x
lands exactly on the four-times guess. Bounded because phase beats cluster rather
than accumulate — 1.0 per push on average, 4 at worst, so the wait after a move is
~2.4s typical.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01X6cF1iYvJ1kNmzYBzu4QX6

CHANGELOG.md

@@ -19,9 +19,295 @@ page as `v0.1.0 · <sha> · <date>`, so what is deployed can always be identifie
 
 ---
 
+## 0.8.0.7 — 2026-09-10
+
+### The Salvage Yard was face up and had nothing to say
+
+*"Why is salvage deck not face up? I should see the card played onto salvage."* It always was — the
+tile reads the top card. It just said **"a card"**.
+
+`apply.ts` pushes a **synthetic id** on `trainScheduled`:
+
+```ts
+s.decks.salvageYard.push(`train-${e.trainNumber}`);
+```
+
+Nothing in `s.cards` matches that, so `cardName()` fell through to its "a card" default — and since a
+train is scheduled several times a Day, that id is on top of the pile most of the time. Measured
+before touching anything: a session's Salvage tile read `"a card"` from the opening frame through 60
+pushes, never once changing, while its depth climbed from 2 to 8.
+
+`cardName()` resolves `train-<n>` now, so the tile reads **Train 3**, **Train 2** as cards land.
+Resolved in `sim/view.ts` because this is a NAME, which is that file's job.
+
+**The engine half is filed, not fixed — Gitea#23.** `reshuffleIfDepleted()` sweeps the Salvage Yard
+back into the draw deck, so a synthetic id can be shuffled in and drawn into a hand as an id with no
+card behind it. Not reachable in ordinary play: eight games driven to 4000 moves across eight seeds
+produced **zero** reshuffles. There are two defensible fixes and the choice turns on what that
+synthetic id is *for*, which is not a call to make in passing while fixing a label.
+
+### Phases scale with the speed control again — at a third of the rate
+
+*"Phases displayed on the upper line go by too quickly still. Should be 4 times as long — at a guess.
+Maybe use the speed multiplier for that too?"*
+
+Two complaints from opposite directions, and the answer is between them. v0.8.0.3 **pinned** phases at
+their tabled beat because scaling them walled off a player's own turn — *"that makes no sense"*. At
+10× that pinned beat is too short to read the sentence on it.
+
+So they scale, damped to a third of the rate: **1× unchanged, 10× lands exactly on the four-times
+guess**, 20× gives 4400ms. The cost stays bounded because phase beats cluster rather than accumulate
+— measured over 60 pushes, a push carries **1.0 phase beat on average and 4 at worst**, so the wait
+after a move is ~2.4s typical and ~10s at its very worst, against the minutes a full multiplier would
+have cost. A player's own move still outlasts a phase beat at every speed, which is the ordering that
+matters.
+
+---
+
+## 0.8.0.6 — 2026-09-10
+
+Playing v0.8.0.5: *"saw bot's office area now — much better."* Three things still wrong, and one of
+them was mine hiding inside the fix for another.
+
+### Your move is put away while the board is catching up
+
+*"Your actions should be hidden while catching up."* Two reasons, and the second is the one that
+changed my mind about a decision taken early in v0.8.0 ("never block input"). The board on screen is
+behind the game, so a move offered there is a move against a position that has already moved on — the
+menu is computed from the CURRENT state and would be acted on while looking at an older one. And the
+screen had grown to four things competing at once: the district, the history, the catching-up row,
+and now a lit pile. Taking the action list out of that competition, while there is nothing to decide
+anyway, is the cheapest way to quieten it.
+
+Not a block: Skip is one click away at the left of the row, so the wait stays voluntary. The buttons
+are replaced by the reason they are gone.
+
+### …which could have locked a player out of their own game
+
+Hiding actions behind `busy()` makes that flag the thing standing between a player and their turn —
+and **without `requestAnimationFrame` nothing ever advances the queue, so `busy()` would never
+clear.** The action list would have been hidden permanently, with Skip the only way to play.
+
+Caught by `test/web.test.ts`, whose DOM stub has no `rAF` — the same stub that has been proving this
+page still starts since long before any of this existed. Two fallbacks now: no `rAF` means draw
+everything at once (exactly what `pace = 0` does deliberately), and a queue that throws empties
+itself rather than stranding the player. `test/step-queue.test.ts` pins that a never-advanced queue
+is still skippable.
+
+### The lit pile was never brief — it was too quiet
+
+*"Never saw decks lighting up… caught one flash deck light up for just a very brief moment, but
+couldn't see that with what bot was doing in office area and history and catch up area all at same
+time."*
+
+Measured before changing anything: at 10× a pile stays lit for **6997ms**, just under seven seconds.
+So the highlight was not brief at all. It was a single 0.45s flash-in over a dark green fill, easy to
+miss entirely while looking at the district — a state that settles stops asking to be looked at. It
+pulses now for as long as the move is up, with a ring and a glow. The reduced-motion fallback is loud
+in a different way rather than simply still, since motion is the whole point here.
+
+### The ceiling was not theoretical
+
+*"At 10× — still a bit fast, but followable."* 10× was the top of the ladder, so the control's
+slowest setting was not slow enough for the person using it. `PACE_LEVELS` now runs to 20 and
+`MAX_PACE` with it. A control whose limit is reached in ordinary use has the wrong limit, not the
+right one held firmly.
+
+---
+
+## 0.8.0.5 — 2026-09-10
+
+**Somewhere to look.** Jesse, playing v0.8.0.4 at 10×: *"many operations still occurred too fast for
+me to see."* At 10× an action holds the screen for seven seconds, so this was never about duration —
+it was that a bot drawing a card changes one number in a panel nobody is watching, and the board sits
+unchanged for those seven seconds. **Raising the dwell was the wrong lever, and it had been pulled
+three times.** His diagnosis was the right one: mark WHERE, not longer.
+
+### The Home Office deck was never drawn
+
+`f.deck` has carried the face-down count since the Frame existed and **nothing in `src/web/` read
+it** — the exact display gap `test/display-gaps.test.ts` was written to sweep for, surviving in the
+one panel that draws every other pile. It is a tile now, first in the row, because that is the order
+a card travels: out of the deck, into a hand, then onto a Department or the Salvage Yard. Face down,
+so its card slot says so rather than naming one — not knowing what is on top is the point of the
+pile.
+
+### What lights, and why that is exactly what is public
+
+The piles a move touched are now lit for as long as that move is on screen. **Derived, never sent**:
+the client already holds the frame before a step and the frame after it, so `changedPiles()` is a
+diff. Nothing is added to the protocol, nothing can drift out of step with the projection, and the
+0.8.1 seatless board gets it for free.
+
+Measured across four seeds rather than reasoned about, and pinned by a test that requires each case
+to have actually occurred rather than passing on whichever the bot happened to play:
+
+| action | lights | why that is public |
+| --- | --- | --- |
+| `draw.fromHomeOffice` | the deck | the count, never the card — a blind draw stays the drawer's |
+| `draw.fromDepartment` | that Department, and the deck when it refills | the pile is face up, so the card taken is public |
+| `card.discard` | that Department | face up, and which pile it went on is the point |
+| `card.play` | the Salvage Yard | where a played card that did not stay on the board lands |
+| switching, new trains | nothing here | they move the board, which the district panel already follows |
+| `*.end`, `localOps.choose` | nothing | no card moved |
+
+**It is a state, not a flash**, and that distinction is the whole reason it works. The timetable's
+existing `.tt-slot.fresh` animates for a fixed 1.5s — right for a die roll nobody is waiting on, and
+wrong here, because a step can hold for seven seconds and the animation would be long over before
+the pause it belongs to. A brief flash-in marks the moment; the lit border and background stay for
+exactly as long as the step is up.
+
+**Not for your own moves.** You drew that card — the same rule that already gives your own steps no
+dwell.
+
+---
+
+## 0.8.0.4 — 2026-09-09
+
+**Housekeeping: the test server's name is out of the ten places this session put it.**
+
+Both of this project's repositories allow anonymous clone — checked, not assumed: `info/refs` for
+`git-upload-pack` answers 200 for `station-master` and for `station-master-startos` alike, while
+`git-receive-pack` answers 401. So everything committed here is public, and the standing rule is that
+tracked files carry placeholders rather than real hosts.
+
+Ten mentions added while building v0.8.0 are now "the test server" or "the target hardware":
+`CHANGELOG.md`, `docs/plans/jitsi-common-board.md` (three identical deferral banners), `sim/pacing.ts`,
+`test/pacing.test.ts` and `test/step-queue.test.ts`. Prose and comments only — no behaviour, and the
+quotes they carry are unchanged, because what a player said about bot pacing is the part worth
+keeping.
+
+**What is deliberately left, and why it is not an oversight:**
+
+- **Nineteen older mentions**, in entries about v0.7.5, v0.7.6 and v0.7.8 and in `TODO.md`. Rewriting
+  a changelog after the fact makes the record less true, and these describe verification that
+  genuinely happened on that machine.
+- **`scripts/deploy-web.ts` is FUNCTIONAL, not prose.** It carries the host as the default for
+  `FB_URL`, so a placeholder there would break the deploy for the person the default exists to serve.
+  Same for the public address it publishes to. If those should move to required environment variables
+  with no default, that is a change to how deploying works and wants deciding on its own rather than
+  being smuggled in beside a comment sweep.
+
+---
+
+## 0.8.0.3 — 2026-09-09
+
+Three things from playing v0.8.0.2, all of them about the row rather than the mechanism.
+
+### Skip was at the wrong end of the row
+
+Jesse: *"the skip button should be on the far left, in front of where it says [the count], so it's
+always close to where people are looking."* It was on the far right, and a player's eye is on the
+countdown. Moved.
+
+### The caption said what, but never who
+
+*"I saw 2 behind, 1 behind, and then it was caught up, but it didn't tell me what the actual action
+was, like who I was waiting on or what they were doing. I knew I was behind, but I wasn't sure what I
+was supposed to be looking for."*
+
+The caption was there. It was the wrong half of the sentence. **Measured over 40 turns of a real
+3-seat game, half the waiting is automatic phases** — 21.0s of phases against 21.7s of other players —
+and a phase narrates as "Mainline", which is accurate and no answer at all to "who am I waiting on".
+A phase now introduces itself: **"The Division: ▸ Mainline phase"**. A player's move already carries
+its name from `record()`, so it is left alone rather than stuttering it twice.
+
+**And the row was hiding a step early.** It was shown only while `behind > 0` — which goes false the
+moment the LAST step of a burst goes up, so the one step a player was most likely to be reading about
+lost its caption. It now stays up while the queue is still showing something, and reads "catching up"
+once nothing is queued behind.
+
+### The speed control was stretching the clock, not just the other players
+
+*"After my turn, when I actually execute my turn, I'm still subject to that same delay before it
+moves on. That makes no sense. Since I've just done my turn, I don't need to wait after it."*
+
+He was right, and it was not his move being replayed — own moves have cost nothing since v0.8.0.1. It
+was the automatic phases behind it, which were scaling with `pace` along with everything else. At 5×
+that put **105 seconds of clock-ticking** into the game, all of it after a player's own move and none
+of it anything to watch.
+
+**`pace` now scales a player's move and leaves a phase at its tabled beat.** The control is labelled
+as how long another player's move is held, and that is now what it does. A phase still gets its beat
+(TODO #18) and still vanishes entirely at `pace = 0`, because off has to mean off.
+
+---
+
+## 0.8.0.2 — 2026-09-09
+
+Two things found by playing v0.8.0.1 on the test server, neither of them in the mechanism itself.
+
+### `?pace=` never worked, and a whole game was played at the wrong speed
+
+Jesse: *"I'm playing at pace = 7, and the bots are still moving too fast for me to follow."* At 7×
+a switching move holds for seven seconds, so that could not be calibration — and it was not. **He was
+at 1× the entire time.**
+
+`index.html`'s two doors are `./play.html?lobby` and `./play.html?solitaire`. Arriving through the
+splash therefore **replaces** the query string, and `location.search` on the play page is `?lobby` —
+so `PACE_OVERRIDE` was null and it fell back to the stored setting of 1. v0.8.0 shipped `?pace=` as
+the only way to change speed and the game's own front door destroyed it. Verified rather than
+assumed: the queue at pace 7 holds a bot's turn for 32.9s with the bot's district up for 24.5s, so
+the mechanism was right and the value never arrived.
+
+Fixed twice over, because one of them is the durable answer:
+
+- **A speed control on the play screen**, beside zoom — `− 1× +`, persisted per viewer, reading
+  through to the queue on the very next move. `PACE_LEVELS` is `0, 0.5, 1, 2, 3, 5, 7, 10`: off is
+  the first rung (TODO #18's "a player who has seen it a hundred times will want it off") and the
+  ladder reaches the speeds people actually reach for. At the top, a six-move switching turn takes a
+  full minute to watch.
+- **The doors now carry `pace` through**, so the URL lever is honest for handing two playtesters
+  different speeds — the only thing it was ever for. When one is present the control says
+  `7× (URL)` and disables itself rather than showing buttons that do nothing.
+
+`PACE_LEVELS` lives in `sim/pacing.ts` with `DWELL` and `MAX_PACE`, not in `main.ts` — the whole
+tuning surface in one file, and testable, which a constant inside the page entry point is not.
+
+**The committed default is unchanged at 1×.** What it should be is a question for a game played at a
+speed that actually took effect.
+
+### "0 today, 2 in all" — the Day-end dialog contradicted itself Jesse, at the end of a Day 1 with
+two collisions in it: *"It shows a total of two collisions, but zero today. Since we just finished day
+one, that does seem to be a contradiction."* Unrelated to v0.8.0 — this has been wrong since the
+dialog was built for Gitea#10, and nobody had played a Day with a collision in it and then read the
+summary.
+
+#### One line of ordering
+
+`advance.ts`, at the rollover:
+
+```ts
+s.clock.day += 1;
+s.collisionsToday = 0;
+```
+
+And `noteDayEnd()` fires when `f.day` goes UP — so the dialog reporting the Day that just finished is
+drawn from the very frame in which that Day's count was zeroed. It printed the *new* Day's zero beside
+a running total that could not possibly agree with it. Reproduced on four of five seeds before
+touching anything: Day 1 ended with `today=3 total=3`, and the dialog read `today=0 total=3`.
+
+**Not derivable on the client, which is why the fix is in the engine.** A Day turns over inside the
+phases that run themselves, so in multiplayer the push announcing the new Day is the same push that
+carries the reset — a client may never see the ended Day's final count to remember it. So
+`collisionsPrevDay` is captured in state at the rollover, immediately before the reset, and rides on
+the frame like the other two counts.
+
+#### And "today" was the wrong word anyway
+
+Even with the right number, a dialog headed "Day 1 has ended" should not say "today" — by then
+"today" is Day 2. It now names the Day: **"Collisions: 2 on Day 1, 2 in all."** The end-of-game
+results screen passes no Day and keeps "today", where the Day has not turned over and the word is
+accurate.
+
+`test/redaction.test.ts`'s allow-list did its job on the way through: adding a public property failed
+the suite until it was declared out loud.
+
+---
+
 ## 0.8.0.1 — 2026-09-09
 
-**Bot play was way too fast.** v0.8.0 was installed on `phoenix.local` and played within the hour;
+**Bot play was way too fast.** v0.8.0 was installed on the test server and played within the hour;
 Jesse: *"I briefly saw that it was the bot's office area then their turn was done and it pointed back
 to my office area"*, and the countdown row appeared "very briefly". Everything else looked right —
 the bots were visibly doing things — so this is calibration and one real bug, not a redesign.

docs/plans/jitsi-common-board.md

@@ -361,14 +361,15 @@ Introduce dedicated allow-listed types. Do not derive them with `Omit<Frame, ...
 > **Read those two, not this**, when building steps 2-7. The differences that matter:
 >
 > - **The shape is FLAT, not grouped.** There is no `clock`, `config`, `scoring` or `deckCounts`
->   object. Their contents sit at the top level. **All 37 properties, which is the same list as
+>   object. Their contents sit at the top level. **All 38 properties, which is the same list as
 >   `test/redaction.test.ts`'s allow-list** — `day`, `stage`, `clock` (a time string), `phase`,
 >   `phaseKey`, `actor`, `superintendent`, `deck`, `departments`, `departmentsWhat`,
 >   `departmentDepth`, `salvage`, `yards`, `timetable`, `timetableWhat`, `houseRules`, `mode`,
 >   `optionalRules`, `days`, `minCombinedRevenue`, `maxCollisionsPerDay`, `maxCollisionsTotal`,
->   `collisionsToday`, `collisionsTotal`, `status`, `outcome`, `extraDays`, `extensionVotes`,
+>   `collisionsToday`, `collisionsPrevDay`, `collisionsTotal`, `status`, `outcome`, `extraDays`,
+>   `extensionVotes`,
 >   `official`, `tally`, `players`, `openingRolls`, `trains`, `crewTrays`, `queued`, `division`,
->   `districts`. The first 35 come from `projectSharedTable`; `division` and `districts` are added
+>   `districts`. The first 36 come from `projectSharedTable`; `division` and `districts` are added
 >   by `PublicFrame` itself.
 > - **`protocolVersion` was NOT built** and exists nowhere in the repo. **Decided 2026-09-09: add it
 >   in step 2.** `display.json` carries its own `schemaVersion`, and the SSE wire format is a second,
@@ -887,7 +888,7 @@ Cover:
 ## Step 5 — Minimal visual-only Jitsi engine
 
 > **v0.9.0 — DEFERRED (2026-09-09).** Steps 5-7 are the Jitsi publisher and are explicitly held
-> off until 0.8.0 ships and Chromium has been measured on `phoenix.local`. Nothing below has been
+> off until 0.8.0 ships and Chromium has been measured on the target hardware. Nothing below has been
 > re-verified against the current code; it was written 2026-08-27. Two things will have moved by
 > the time it is picked up: the control channel is **SSE + POST, not a WebSocket** (§ THE RELEASE
 > SPLIT), and adding Chromium takes the `.s9pk` from 63 MB to several hundred against a package
@@ -1027,7 +1028,7 @@ Port/adapt the sibling repository’s proven tests for:
 ## Step 6 — Chromium publisher supervisor
 
 > **v0.9.0 — DEFERRED (2026-09-09).** Steps 5-7 are the Jitsi publisher and are explicitly held
-> off until 0.8.0 ships and Chromium has been measured on `phoenix.local`. Nothing below has been
+> off until 0.8.0 ships and Chromium has been measured on the target hardware. Nothing below has been
 > re-verified against the current code; it was written 2026-08-27. Two things will have moved by
 > the time it is picked up: the control channel is **SSE + POST, not a WebSocket** (§ THE RELEASE
 > SPLIT), and adding Chromium takes the `.s9pk` from 63 MB to several hundred against a package
@@ -1160,7 +1161,7 @@ Use fake child processes and fake control sockets to test:
 ## Step 7 — Configuration, lifecycle, packaging, and observability
 
 > **v0.9.0 — DEFERRED (2026-09-09).** Steps 5-7 are the Jitsi publisher and are explicitly held
-> off until 0.8.0 ships and Chromium has been measured on `phoenix.local`. Nothing below has been
+> off until 0.8.0 ships and Chromium has been measured on the target hardware. Nothing below has been
 > re-verified against the current code; it was written 2026-08-27. Two things will have moved by
 > the time it is picked up: the control channel is **SSE + POST, not a WebSocket** (§ THE RELEASE
 > SPLIT), and adding Chromium takes the `.s9pk` from 63 MB to several hundred against a package

package.json

@@ -1,6 +1,6 @@
 {
   "name": "station-master",
-  "version": "0.8.0.1",
+  "version": "0.8.0.7",
   "private": true,
   "type": "module",
   "description": "Station Master — a railroad operations game",

src/engine/advance.ts

@@ -1638,6 +1638,9 @@ function shiftChange(s: GameState, events: GameEvent[]): AdvanceResult {
 
     s.clock.day += 1;
     s.clock.stage = 1;
+    // Captured BEFORE the reset: the Day-end dialog reports the Day that just finished, and it is
+    // drawn from the frame this rollover produces. See `collisionsPrevDay` in `state.ts`.
+    s.collisionsPrevDay = s.collisionsToday;
     s.collisionsToday = 0;
     events.push({ type: 'stageBegan', day: s.clock.day, stage: 1 });
     rotateSeats(s, events);

src/engine/setup.ts

@@ -431,6 +431,7 @@ export function createGame(opts: SetupOptions): GameState {
     turns: freshTurns(playerCount, MOVES_PER_LOCAL_OPS),
     movedThisPhase: new Set(),
     collisionsToday: 0,
+    collisionsPrevDay: 0,
     collisionsTotal: 0,
     status: 'active',
     outcome: null,

src/engine/state.ts

@@ -1119,6 +1119,20 @@ export type GameState = {
   movedThisPhase: Set<TrayId>;
   /** §3.4 — resets at the start of each Day; checked against `config.maxCollisionsPerDay`. */
   collisionsToday: number;
+  /**
+   * What `collisionsToday` held for the Day that just ENDED — captured at the rollover, immediately
+   * before the reset.
+   *
+   * The Day-end dialog exists to report the Day that finished, and it is drawn from the frame AFTER
+   * the rollover, because that is the frame whose `day` went up. So it read `collisionsToday` as 0 no
+   * matter what had happened: Jesse, 2026-09-09, at the end of a Day 1 with two collisions in it —
+   * "it shows a total of two collisions, but zero today ... that does seem to be a contradiction".
+   *
+   * NOT DERIVABLE ON THE CLIENT. A Day turns over inside the phases that run themselves, so in
+   * multiplayer the push that reports the new Day is the same push that reports the reset — a client
+   * may never see the ended Day's final count to remember it.
+   */
+  collisionsPrevDay: number;
   /** §3.4 — never reset; checked against `config.maxCollisionsTotal`. */
   collisionsTotal: number;
   /**

src/sim/pacing.ts

@@ -54,7 +54,7 @@ export const DWELL: Record<StepKind, number> = {
    *
    * WAS 250ms, WHICH WAS WRONG, and wrong in the way that mattered most: an early-game bot turn has
    * no switching in it at all, so it was six steps of 250ms and 0ms — **750ms for a whole turn**.
-   * Jesse, from the first real play on `phoenix.local`: *"bot play was way too fast. I briefly saw
+   * Jesse, from the first real play on the test server: *"bot play was way too fast. I briefly saw
    * that it was the bot's office area then their turn was done."* His instruction had been "start at
    * 1s and tune down", and that was applied only to switching while this number was invented.
    */
@@ -143,10 +143,27 @@ export function kindOf(cause: StepCause): StepKind {
  *
  * `pace` has no lower surprise — 0 means off — but an unbounded upper one does: `?pace=300` from
  * somebody typing 3.00, or a corrupt `localStorage` value, would give a switching move a five-minute
- * dwell and look exactly like a frozen board. Ten is far beyond any speed anyone would choose (2 and
- * 3 are the ones actually asked for) and well short of unusable.
+ * dwell and look exactly like a frozen board. Twenty is far past any speed anyone would choose and
+ * well short of unusable.
+ *
+ * RAISED FROM TEN 2026-09-10, because the ceiling turned out not to be theoretical: Jesse played at
+ * 10× — the top of the ladder — and reported it *"still a bit fast, but followable"*. A control whose
+ * slowest setting is not slow enough for the person using it has the wrong ceiling, not the right one
+ * held firmly.
  */
-export const MAX_PACE = 10;
+export const MAX_PACE = 20;
+
+/**
+ * The speeds the on-screen control offers, slowest last.
+ *
+ * `0` is off: every move is drawn at once, as it was before v0.8.0 — TODO #18's "a player who has
+ * seen it a hundred times will want it off". The ladder runs well past 1 because that is what the
+ * first real play asked for: Jesse reached for 7×, and although the `?pace=` he used never took
+ * effect (the splash replaces the query string, so the play page only ever saw `?lobby`), the wish
+ * was real. Watching a bot shunt cars is the point of this feature, and it is worth as long as it
+ * takes.
+ */
+export const PACE_LEVELS = [0, 0.5, 1, 2, 3, 5, 7, 10, 15, 20] as const;
 
 /**
  * How long to show one step, in ms, at a given speed.
@@ -160,6 +177,28 @@ export function dwellFor(cause: StepCause, pace = 1): number {
   return Math.round(DWELL[kindOf(cause)] * Math.min(MAX_PACE, Math.max(0, pace)));
 }
 
+/**
+ * How much of the speed control a PHASE gets — damped, not the full multiplier.
+ *
+ * Phases were pinned at their tabled beat in v0.8.0.3, because scaling them with everything else put
+ * a wall of clock-ticking after a player's own move. That was right about the cost and wrong about
+ * the need: at 10× the caption row goes past faster than the sentence on it can be read. Jesse,
+ * 2026-09-10: *"phases displayed on the upper line go by too quickly still. Should be 4 times as
+ * long — at a guess. Maybe use the speed multiplier for that too?"*
+ *
+ * So they scale, at a third of the rate. That lands exactly on his guess — 10× gives a phase four
+ * times its tabled beat — while leaving 1× untouched, and it stays affordable because phase beats
+ * cluster rather than accumulate: measured over 60 pushes, a push carries **1.0 phase beat on
+ * average and 4 at worst**, so the wait after a move goes to ~2.4s typical and ~10s at its very
+ * worst rather than the minutes a full multiplier would have cost.
+ *
+ * Below 1× it simply follows the multiplier: somebody asking for everything faster means the phases
+ * too.
+ */
+function phaseSpeed(pace: number): number {
+  return pace <= 1 ? pace : 1 + (pace - 1) / 3;
+}
+
 /**
  * How long to show one STEP — the form the queue actually uses.
  *
@@ -170,10 +209,25 @@ export function dwellFor(cause: StepCause, pace = 1): number {
  * have to import `DisplayStep` back from the module that imports `StepCause` from it.
  */
 export function dwellForStep(
-  step: { cause: StepCause; lines: readonly unknown[]; frame: { table: object } },
+  step: { cause: StepCause; player: number | null; lines: readonly unknown[]; frame: { table: object } },
   pace = 1,
 ): number {
-  if (step.lines.length > 0) return dwellFor(step.cause, pace);
+  // Off means off, for the clock as much as for anybody's move.
+  if (pace <= 0) return 0;
+  /**
+   * THE SPEED CONTROL IS ABOUT OTHER PEOPLE, NOT ABOUT THE CLOCK.
+   *
+   * A phase keeps its tabled beat at every speed. Measured over 40 turns of a real 3-seat game, the
+   * waiting split almost evenly — 21.0s of other players against 21.0s of phases turning over — so
+   * scaling both put 105 seconds of clock-ticking into a 5× game, all of it after the player's own
+   * move and none of it anything to watch. Jesse, from that game: *"after my turn, when I actually
+   * execute my turn, I'm still subject to that same delay before it moves on. That makes no sense."*
+   *
+   * The phase still gets its beat (TODO #18) — it just does not get longer because somebody wanted
+   * to watch a bot shunt cars.
+   */
+  const speed = step.player === null ? phaseSpeed(pace) : pace;
+  if (step.lines.length > 0) return dwellFor(step.cause, speed);
   /**
    * A SILENT STEP EARNS A BEAT ONLY WHEN THE CLOCK TURNED OVER — which is TODO #18 exactly: "give
    * every phase a visible beat", for New Train, the Mainline and the shift change.
@@ -182,12 +236,11 @@ export function dwellForStep(
    * silently killed #18: a phase can move trains without saying anything, and those steps were being
    * flashed past. "Anything that changed the board" is wrong the other way: `submit()` steps
    * `advance()` about 4.6 times per intent and most of those merely hand the turn on, so beating on
-   * all of them would cost a quarter of an hour a game. The phase turning over is the thing a player
-   * is being shown, and there are about 180 of those in a full game.
+   * all of them would cost a quarter of an hour a game.
    */
   const table = step.frame.table as Record<string, unknown>;
   const turned = 'phase' in table || 'phaseKey' in table || 'stage' in table || 'day' in table;
-  return turned ? dwellFor(step.cause, pace) : 0;
+  return turned ? dwellFor(step.cause, speed) : 0;
 }
 
 /**

src/sim/public-delta.ts

@@ -130,3 +130,46 @@ function need<T>(value: T | undefined, what: string): T {
   }
   return value;
 }
+
+/** A face-up or face-down pile a card can move to or from, as the display addresses it. */
+export type PileKey = 'home' | 'salvage' | `dept${number}`;
+
+/**
+ * WHICH PILES A STEP MOVED — derived, never sent.
+ *
+ * The receiver already holds the frame before a step and the frame after it, so which pile changed
+ * is a diff rather than something the wire has to carry. That matters twice over: nothing is added
+ * to the protocol, and it cannot drift out of step with the projection the way a hand-maintained
+ * hint would.
+ *
+ * WHY IT IS NEEDED AT ALL. A player watching somebody else draw a card sees seven seconds of an
+ * unchanged board — the step holds the screen, and the only thing that moved is a number in a panel
+ * they were not looking at. Jesse, playing v0.8.0.4 at 10×: *"many operations still occurred too fast
+ * for me to see"*, which was never about duration. Lighting the pile is what tells the eye where.
+ *
+ * WHAT EACH ACTION MOVES, measured across four seeds rather than reasoned about:
+ *
+ * | intent                  | piles                                                    |
+ * | ----------------------- | -------------------------------------------------------- |
+ * | `draw.fromHomeOffice`   | `home` — the COUNT only; the card itself stays private    |
+ * | `draw.fromDepartment`   | that `dept`, and `home` too when the pile refills from it |
+ * | `card.discard`          | that `dept`                                               |
+ * | `card.play`             | `salvage`, or nothing here when it lands on the board      |
+ * | switching, new trains   | nothing here — those show on the board itself             |
+ */
+export function changedPiles(before: PublicFrame | null, after: PublicFrame): PileKey[] {
+  if (before === null) return [];
+  const out: PileKey[] = [];
+  if (before.deck !== after.deck) out.push('home');
+  after.departmentDepth.forEach((depth, i) => {
+    // The TOP as well as the depth: taking the face-up card and replacing it leaves the count alone
+    // and changes the card everybody can see, which is the half that matters to a watcher.
+    if (before.departmentDepth[i] !== depth || before.departments[i] !== after.departments[i]) {
+      out.push(`dept${i}`);
+    }
+  });
+  if (before.salvage.depth !== after.salvage.depth || before.salvage.top !== after.salvage.top) {
+    out.push('salvage');
+  }
+  return out;
+}

src/sim/view.ts

@@ -437,6 +437,8 @@ export type Frame = {
   maxCollisionsPerDay: number;
   maxCollisionsTotal: number;
   collisionsToday: number;
+  /** What the Day that just ended finished on — see `collisionsPrevDay` in `engine/state.ts`. */
+  collisionsPrevDay: number;
   collisionsTotal: number;
   status: GameState['status'];
   outcome: GameState['outcome'];
@@ -1594,6 +1596,7 @@ export function projectSharedTable(s: GameState) {
     maxCollisionsPerDay: s.config.maxCollisionsPerDay,
     maxCollisionsTotal: s.config.maxCollisionsTotal,
     collisionsToday: s.collisionsToday,
+    collisionsPrevDay: s.collisionsPrevDay,
     collisionsTotal: s.collisionsTotal,
     status: s.status,
     outcome: s.outcome,
@@ -1788,6 +1791,23 @@ export function snapshot(
 
 /** A card id turned into something a person can read. */
 export function cardName(s: GameState, id: string): string {
+  /**
+   * A SCHEDULED TRAIN IS IN THE SALVAGE YARD UNDER A SYNTHETIC ID, and without this the pile that is
+   * supposed to be face up reads "a card".
+   *
+   * `trainScheduled` pushes `train-<number>` rather than the id of the card that was played
+   * (`apply.ts`), so there is nothing in `s.cards` to look up — and since a train is scheduled
+   * several times a Day, that synthetic id is on top of the Salvage Yard most of the time. Reported
+   * by Jesse 2026-09-10 as "why is salvage deck not face up. I should see the card played onto
+   * salvage": it was face up all along and simply had nothing to say.
+   *
+   * Resolved here rather than in the engine because this is a NAME, which is this file's job. Whether
+   * the engine should be pushing a real card id instead is a separate question with a separate
+   * consequence — the Salvage Yard is swept back into the draw deck when it runs out — and is filed
+   * rather than answered in passing.
+   */
+  const scheduled = /^train-(\d+)$/.exec(id);
+  if (scheduled) return `Train ${scheduled[1]}`;
   const k = s.cards.get(id)?.kind;
   if (!k) return 'a card';
   switch (k.kind) {

src/web/main.ts

@@ -26,6 +26,7 @@ import { createLocalSession, createRemoteSession } from './session.ts';
 import type { PlayerIndex } from '../engine/state.ts';
 import type { PublicDistrict } from '../sim/view.ts';
 import { createStepQueue } from './step-queue.ts';
+import { PACE_LEVELS } from '../sim/pacing.ts';
 import { notice, prefillCode, runLobby } from './lobby.ts';
 import type { LobbyReady } from './lobby.ts';
 import {
@@ -53,6 +54,7 @@ const REMOTE_KEY = 'station-master.remote.v1';
 /** Preset board zoom levels — a fraction applied to the rendered SVG's own pixel dimensions. */
 const ZOOM_LEVELS = [0.75, 1, 1.25, 1.5] as const;
 
+
 /**
  * Small persisted preferences, kept in a `localStorage` key of their own — separate from
  * `SAVE_KEY`. A save is the seed plus the intents and has to stay portable; none of this belongs
@@ -182,6 +184,20 @@ function drainIntoQueue(): void {
   const reset = session.takeDisplayReset();
   if (reset) stepQueue.reset(reset);
   stepQueue.push(session.takeDisplaySteps());
+  /**
+   * NOWHERE TO ANIMATE MEANS DO NOT QUEUE AT ALL.
+   *
+   * Without `requestAnimationFrame` nothing ever advances the queue, so `busy()` would stay true for
+   * good — and since "Your Move" is now put away while the board is catching up, that would hide a
+   * player's own actions permanently, leaving Skip as the only way to play the game. Drawing
+   * everything at once is exactly what `pace = 0` does deliberately, so that is the honest fallback
+   * rather than a broken page. Caught by `test/web.test.ts`, whose DOM stub has no `rAF` — the same
+   * stub that has been proving this page still starts since long before any of this existed.
+   */
+  if (typeof requestAnimationFrame !== 'function') {
+    stepQueue.skip();
+    return;
+  }
   if (stepQueue.busy()) startAnimationLoop();
 }
 
@@ -199,17 +215,23 @@ function drainIntoQueue(): void {
  * drawn — from `PublicDistrict`, which `officeSvg` can render as-is because it takes board data and
  * has never needed a private viewer.
  */
-function renderWatching(): void {
+function renderWatching(f?: Frame): void {
   const behind = stepQueue.behind();
   const row = $('watching');
-  // Collapsed whenever the board is level with the game — which in solitaire is nearly always, and
-  // between turns in multiplayer too. A row that is always there would be a row nobody reads.
-  if (behind === 0) {
+  /**
+   * VISIBLE WHILE THE BOARD IS BEHIND **OR** STILL SHOWING SOMETHING.
+   *
+   * It used to hide the moment `behind` hit zero — which is the moment the LAST step of a burst goes
+   * up, so the one step a player was most likely to be reading about lost its caption. Collapsed
+   * otherwise: in solitaire that is nearly always, and between turns in multiplayer too, and a row
+   * that is always there is a row nobody reads.
+   */
+  if (behind === 0 && !stepQueue.busy()) {
     row.hidden = true;
     return;
   }
   row.hidden = false;
-  $('watching-behind').textContent = `${behind} behind`;
+  $('watching-behind').textContent = behind === 0 ? 'catching up' : `${behind} behind`;
   /**
    * THE CAPTION IS #15, and this is where that item lands rather than as a line of its own.
    *
@@ -218,8 +240,29 @@ function renderWatching(): void {
    * The queue IS that, so the caption simply names the step being shown, and the counter beside it
    * says how much of the wait is left.
    */
+  /**
+   * WHO, THEN WHAT — Jesse, 2026-09-09: *"it didn't tell me what the actual action was, like who I
+   * was waiting on or what they were doing. I knew I was behind, but I wasn't sure what I was
+   * supposed to be looking for."*
+   *
+   * The caption was there; it was the wrong half of the sentence. Half the waiting is automatic
+   * phases, whose narration reads "Mainline" — accurate, and no answer at all to "who am I waiting
+   * on". So the name goes first, and a phase says so in as many words rather than leaving the reader
+   * to infer that nobody is acting.
+   *
+   * The narrated line is used as it stands otherwise, because `record()` already prefixes it with the
+   * player — "Player Bot 1 moved Train 3 (−1,−2) → (−1,1)" — so a second name would stutter.
+   */
   const showing = stepQueue.showing();
-  $('watching-what').textContent = showing?.lines[0]?.text ?? '';
+  const said = showing?.lines[0]?.text ?? '';
+  const who =
+    showing === null || showing === undefined
+      ? ''
+      : showing.player === null
+        ? 'The Division'
+        : (f?.players[showing.player]?.name ?? `Seat ${seatLabel(showing.player)}`);
+  // A player action already names its actor; a phase does not, so it is introduced.
+  $('watching-what').textContent = showing?.player === null && said !== '' ? `${who}: ${said}` : said;
   /**
    * SKIP COSTS THE ANIMATION AND NEVER THE INFORMATION.
    *
@@ -287,12 +330,25 @@ function startAnimationLoop(): void {
        */
       console.error('display queue stopped:', err);
       animating = false;
+      // Do not strand the player behind a queue that can no longer advance: jump the board to the
+      // live position, which brings "Your Move" back with it.
+      try {
+        stepQueue.skip();
+      } catch {
+        /* nothing further to try — the authoritative Frame is still what the rest of the page draws */
+      }
+      render();
       return;
     }
     if (!stepQueue.busy()) {
       animating = false;
-      // One last render so the "N behind" row collapses the moment the board is level.
-      renderWatching();
+      /**
+       * A FULL RENDER, not just the row. The board being level again is what brings "Your Move"
+       * back and clears the last lit pile, so redrawing only the catching-up row would leave the
+       * action list hidden until something else happened to trigger a render — which, when the game
+       * is waiting on this player, is nothing at all.
+       */
+      render();
       return;
     }
     requestAnimationFrame(tick);
@@ -1262,7 +1318,7 @@ function render(): void {
 
   renderTurnChart(f);
   renderPresence(f);
-  renderWatching();
+  renderWatching(f);
   $('revenue').textContent = String(f.revenue);
   /**
    * THE OBJECTIVE, WITHOUT THE COMMENTARY.
@@ -1482,7 +1538,9 @@ function render(): void {
    * reach. Drawn like the hand so they read as cards, dashed and unlit because taking one is a draw
    * action rather than a click on the card itself.
    */
-  $('depts').innerHTML = pilesHtml(f);
+  // The pile the move being WATCHED just touched, lit for as long as that step is on screen. Empty
+  // whenever the board is level with the game, or when the move was this player's own.
+  $('depts').innerHTML = pilesHtml(f, stepQueue.busy() ? stepQueue.lit() : []);
 
   renderYards(f);
 
@@ -1956,6 +2014,29 @@ function renderActions(
     renderEnding(el, f);
     return;
   }
+
+  /**
+   * YOUR MOVE IS PUT AWAY WHILE THE BOARD IS CATCHING UP — Jesse, 2026-09-10: *"your actions should
+   * be hidden while catching up."*
+   *
+   * Two reasons, and the second is the one that changed my mind about it. The board on screen is
+   * behind the game, so a move offered here is a move against a position that has already moved on —
+   * the menu is computed from the CURRENT state and would be acted on while looking at an older one.
+   * And the display had grown to four things demanding attention at once — the district, the history,
+   * the catching-up row and now a lit pile — which is what made the pile highlight so easy to miss.
+   * Taking the action list out of that competition while there is nothing to decide anyway is the
+   * cheapest way to quieten it.
+   *
+   * NOT A BLOCK. Skip is one click away and sits at the left of the row, so the wait is always
+   * voluntary; this replaces the buttons with the reason they are gone, rather than leaving a live
+   * menu over a stale board.
+   */
+  if (stepQueue.busy()) {
+    el.innerHTML =
+      '<div class="dim">Catching up on what everyone else did — your move is here when the board is ' +
+      'level with the game. <b>Skip</b> jumps straight to it.</div>';
+    return;
+  }
   // The game is running, so the next ending — an extended Day's, or a fresh game's — is entitled to
   // put its results up unasked again (Gitea#11).
   resultsShown = false;
@@ -2603,6 +2684,52 @@ function runSolitaireSetup(params: URLSearchParams, hasSave = false, live: Frame
   dealBtn.onclick = () => commitNewGame(ss, seedField?.value ?? '');
 }
 
+/**
+ * PLAYBACK SPEED — v0.8.0.3, TODO #13.
+ *
+ * Persisted per viewer in `Settings`, so it survives the navigation that was eating `?pace=`. The
+ * queue reads `settings.pace` through a closure on every step, so a change here takes effect on the
+ * very next move rather than the next game.
+ */
+const paceSlowerBtn = document.getElementById('paceslower') as HTMLButtonElement | null;
+const paceFasterBtn = document.getElementById('pacefaster') as HTMLButtonElement | null;
+const paceLabel = document.getElementById('pacelabel');
+if (paceSlowerBtn && paceFasterBtn && paceLabel) {
+  const nearestPace = (): number => {
+    // A saved or URL value need not be on the ladder — `?pace=7` and a hand-edited setting are both
+    // legitimate — so the buttons step from whichever preset is closest rather than refusing to move.
+    const want = PACE_OVERRIDE ?? settings.pace;
+    return PACE_LEVELS.reduce((best, p) => (Math.abs(p - want) < Math.abs(best - want) ? p : best), PACE_LEVELS[0]);
+  };
+  const paintPace = (): void => {
+    const p = PACE_OVERRIDE ?? settings.pace;
+    paceLabel.textContent = p === 0 ? 'off' : `${p}×`;
+    const i = PACE_LEVELS.indexOf(nearestPace() as (typeof PACE_LEVELS)[number]);
+    paceSlowerBtn.disabled = i >= PACE_LEVELS.length - 1;
+    paceFasterBtn.disabled = i <= 0;
+    // A `?pace=` in the URL wins over the setting, so say so rather than showing dead buttons.
+    if (PACE_OVERRIDE !== null) {
+      paceSlowerBtn.disabled = true;
+      paceFasterBtn.disabled = true;
+      paceLabel.textContent = `${PACE_OVERRIDE}× (URL)`;
+    }
+  };
+  const stepPace = (by: number): void => {
+    const i = PACE_LEVELS.indexOf(nearestPace() as (typeof PACE_LEVELS)[number]);
+    const next = PACE_LEVELS[Math.min(PACE_LEVELS.length - 1, Math.max(0, i + by))];
+    if (next === undefined) return;
+    saveSettings({ pace: next });
+    paintPace();
+    // The row's countdown is measured in steps that will dwell, so a change to 0 empties it at once.
+    renderWatching();
+  };
+  // Slower is a BIGGER multiplier, so "−" walks up the ladder. Labelled by what it does to the game,
+  // not to the number: a player pressing "slower" wants to watch for longer.
+  paceSlowerBtn.onclick = () => stepPace(1);
+  paceFasterBtn.onclick = () => stepPace(-1);
+  paintPace();
+}
+
 const zoomOutBtn = document.getElementById('zoomout') as HTMLButtonElement | null;
 const zoomInBtn = document.getElementById('zoomin') as HTMLButtonElement | null;
 const zoomLabel = document.getElementById('zoomlabel');

src/web/panels.ts

@@ -52,25 +52,61 @@ export function handHtml(f: Frame, canPlay: (boolean | null)[] = []): string {
  * Only the top card may ever be drawn, so the depth is a count and not a hint: everything below it
  * is out of reach, and choosing where to discard is choosing what to put there.
  */
-export function pilesHtml(f: Frame): string {
-  const pile = (label: string, top: string, depth: number, why: string, extra = '', slot = -1): string => {
+export function pilesHtml(f: Frame, lit: readonly string[] = []): string {
+  const pile = (
+    key: string,
+    label: string,
+    top: string,
+    depth: number,
+    why: string,
+    extra = '',
+    slot = -1,
+    faceDown = false,
+  ): string => {
     const tip = [why, extra].filter(Boolean).join(' · ');
     // A Department is a DROP TARGET for a discard. The attribute is always emitted; only the play
     // page binds a click to it, and only while a card is waiting to be discarded — so the replay
     // viewer draws exactly the same markup and nothing there is clickable.
     const target = slot >= 0 ? ` data-dept="${slot}"` : '';
+    // `lit` marks the pile the move being watched just touched — see `changedPiles`.
+    const cls = `handcard${faceDown ? ' facedown' : ''}${lit.includes(key) ? ' pilelit' : ''}`;
     return (
-      `<div class="handcard"${target}${tip ? ` data-tip="${esc(tip)}"` : ''} tabindex="0">` +
+      `<div class="${cls}"${target}${tip ? ` data-tip="${esc(tip)}"` : ''} tabindex="0">` +
       `<div class="pilehd"><span>${esc(label)}</span><span class="depth">${depth}</span></div>` +
       `<b>${esc(top)}</b></div>`
     );
   };
   return (
+    /**
+     * THE HOME OFFICE DECK, which the screen had never drawn.
+     *
+     * `f.deck` has carried the face-down count since the Frame existed and nothing read it — the
+     * exact shape of display gap `test/display-gaps.test.ts` was written to sweep for, surviving in
+     * the panel that draws every OTHER pile. Asked for by Jesse 2026-09-10 for a second reason: a
+     * player drawing from it is the commonest move nobody can see, so it needs somewhere to flash.
+     *
+     * FIRST, because that is the order a card travels: out of here, into a hand, then onto a
+     * Department or the Salvage Yard. Face down, so the card slot says so rather than naming a card
+     * — the whole point of this pile is that nobody knows what is on top.
+     */
+    pile(
+      'home',
+      'Home Office',
+      'face down',
+      f.deck,
+      'The draw deck. Face down — nobody sees what is on top, and a card drawn from here is private ' +
+        'to whoever drew it. When it runs out, the Salvage Yard and the Departments are swept back ' +
+        'into it.',
+      '',
+      -1,
+      true,
+    ) +
     f.departments
       .map((d, i) => {
         const depth = f.departmentDepth[i] ?? 0;
         const under = depth - 1;
         return pile(
+          `dept${i}`,
           `Dept ${i + 1}`,
           d,
           depth,
@@ -81,6 +117,7 @@ export function pilesHtml(f: Frame): string {
       })
       .join('') +
     pile(
+      'salvage',
       'Salvage',
       f.salvage.top,
       f.salvage.depth,
@@ -180,7 +217,7 @@ export function dayEndHtml(f: Frame): string {
     ahead +
     standingsHtml(f) +
     targetHtml(f) +
-    collisionsHtml(f)
+    collisionsHtml(f, ended)
   );
 }
 
@@ -238,13 +275,28 @@ function targetHtml(f: Frame): string {
  * its config and enforces neither, so reporting a collision budget there would put a rule on
  * screen that this game does not have.
  */
-function collisionsHtml(f: Frame): string {
+function collisionsHtml(f: Frame, endedDay?: number): string {
   const scoredOnCollisions =
     (f.mode === 'competitive' || f.mode === 'coop') &&
     (f.maxCollisionsTotal > 0 || f.maxCollisionsPerDay > 0);
-  return scoredOnCollisions
-    ? `<p>Collisions: <b>${f.collisionsToday}</b> today, <b>${f.collisionsTotal}</b> in all.</p>`
-    : '';
+  if (!scoredOnCollisions) return '';
+  /**
+   * "TODAY" IS THE WRONG WORD IN A DAY-END DIALOG, and it read as a contradiction.
+   *
+   * That dialog is drawn from the frame whose `day` went UP — which is the same frame in which
+   * `collisionsToday` was reset — so it reported 0 however many there had been. Jesse, 2026-09-09,
+   * at the end of a Day 1 with two collisions in it: "it shows a total of two collisions, but zero
+   * today ... that does seem to be a contradiction."
+   *
+   * So when the caller knows which Day just ended it says so by name, and reads the count captured at
+   * the rollover. The end-of-game results screen passes nothing and keeps "today", where the Day has
+   * not turned over and the word is accurate.
+   */
+  const [count, when] =
+    endedDay === undefined
+      ? [f.collisionsToday, 'today']
+      : [f.collisionsPrevDay, `on Day ${endedDay}`];
+  return `<p>Collisions: <b>${count}</b> ${when}, <b>${f.collisionsTotal}</b> in all.</p>`;
 }
 
 /**
@@ -661,6 +713,30 @@ h3{font-size:11px;text-transform:uppercase;letter-spacing:.07em;color:#8b94a3;ma
 .handcard:focus{outline:2px solid #4d6fa8;outline-offset:1px}
 .cardrow.ref .handcard{background:#1c2129;border-style:dashed;border-color:#39424e;color:#b6bec9}
 .handcard.unplayable{color:#7d8794;border-color:#39424e}
+/* THE HOME OFFICE DECK. Face down, so its card slot names no card — it says so instead, in the
+   dimmed voice the rest of the panel uses for "nothing to read here". */
+.handcard.facedown > b{color:#6f7885;font-style:italic;font-weight:400}
+/* THE PILE A WATCHED MOVE JUST TOUCHED (v0.8.1).
+   A STATE, NOT A FLASH, and that is the whole point. The .tt-slot.fresh rule above animates for a fixed
+   1.5s, which is right for a die roll nobody is waiting on — but a step can hold the screen for
+   seven seconds at 10x, so a fixed animation would be over long before the pause it belongs to and
+   the player would be back to staring at an unchanged board. The flash-in marks the moment; the lit
+   border and background stay for exactly as long as the step is up, because the class is on the
+   element only while that step is the one being shown. */
+.handcard.pilelit{border-color:#8fd6a0;background:#1d3327;box-shadow:0 0 0 2px #2f6b47,0 0 14px rgba(143,214,160,.55);
+  animation:pilepulse 1.15s ease-in-out infinite}
+/* A PULSE FOR THE WHOLE DWELL, not one flash at the start. Measured: at 10x a pile stays lit for
+   just under seven seconds, so the highlight was never brief — but a single 0.45s flash-in and a
+   dark green fill were easy to miss entirely while watching the district. Jesse: "caught one flash
+   deck light up for just a very brief moment, but couldn't see that with what bot was doing in
+   office area and history and catch up area all at same time." Something still moving keeps drawing
+   the eye for as long as the move is up; a state that settles stops asking to be looked at. */
+@keyframes pilepulse{0%,100%{background:#1d3327;box-shadow:0 0 0 2px #2f6b47,0 0 14px rgba(143,214,160,.45)}
+  50%{background:#2f6b47;box-shadow:0 0 0 3px #8fd6a0,0 0 22px rgba(143,214,160,.85)}}
+/* Motion is the point here, so the reduced-motion fallback has to be loud in a different way rather
+   than simply not moving: a solid ring and a brighter fill, held. */
+@media(prefers-reduced-motion:reduce){
+  .handcard.pilelit{animation:none;background:#2f6b47;box-shadow:0 0 0 3px #8fd6a0}}
 .handcard.unplayable::after{content:"";position:absolute;inset:0;border-radius:5px;pointer-events:none;
   background:repeating-linear-gradient(45deg,transparent 0 5px,rgba(150,160,175,.20) 5px 6px)}
 /* THE CARD JUST DRAWN. It sits first in the row, and this says which one that is — three cards that

src/web/play.html

@@ -125,6 +125,7 @@ section{background:var(--panel);border:1px solid var(--line);border-radius:7px;
 .wbehind{font-variant-numeric:tabular-nums;font-weight:700;color:#c9cee0;
   background:#22263a;border:1px solid #343a52;border-radius:10px;padding:1px 8px;white-space:nowrap}
 #watching-skip{font-size:11px;padding:2px 10px;border-color:#e0b060;color:#e0b060;flex:none}
+#watching-who{color:#c9cee0;font-weight:700}
 #presence:empty{display:none}
 /* division strip */
 #division{display:flex;gap:7px;overflow-x:auto;padding-bottom:4px}
@@ -872,6 +873,14 @@ ul.blocked li{padding:2px 0}
   <span class="zoom" title="Zoom the Division map and your Office Area. Both already scroll — this only changes their size.">
     <button id="zoomout" aria-label="Zoom out">−</button><span id="zoomlabel">100%</span><button id="zoomin" aria-label="Zoom in">+</button>
   </span>
+  <!-- HOW FAST OTHER PLAYERS' TURNS PLAY BACK — v0.8.0.3, TODO #13.
+       A CONTROL RATHER THAN ONLY A URL PARAMETER. `?pace=` shipped first and is unreachable through
+       the front door: `index.html`'s two doors are `play.html?lobby` and `play.html?solitaire`, so
+       arriving from the splash REPLACES the query string and any pace with it. Jesse played a whole
+       game believing he was at 7x when he was at 1x. -->
+  <span class="zoom" title="How long another player's or a bot's move is held on screen before the next one. Yours are never delayed. Off draws every move at once, as it did before v0.8.0.">
+    <button id="paceslower" aria-label="Slower playback">−</button><span id="pacelabel">1×</span><button id="pacefaster" aria-label="Faster playback">+</button>
+  </span>
   <button id="undo" title="Take the last action back. The save is the seed plus the moves made, so this replays the game without the last one — as far back as you like.">Undo</button>
   <button id="savefile" title="Download this game as a save file you can replay or share">Save replay</button>
   <button id="newgame" title="Set up a fresh game — the seed, the table, the opening hand and what the three economies pay. Opens the same screen a new solitaire game starts from, with your current rules filled in; your game in progress is kept until you press Deal, and Continue puts it straight back.">New game</button>
@@ -906,9 +915,12 @@ ul.blocked li{padding:2px 0}
      and Skip. Empty and collapsed whenever the board is level with the game, which in solitaire is
      almost always. -->
 <div id="watching" hidden>
+  <!-- SKIP FIRST, on the left. It sat on the far right and a player's eye is on the countdown, not at
+       the other end of the row — Jesse, 2026-09-09: "the skip button should be on the far left, in
+       front of where it says [the count], so it's always close to where people are looking." -->
+  <button id="watching-skip" class="ghost" type="button" title="Stop animating and jump the board to where the game actually is. Nothing is lost — every line is already in the History panel.">Skip</button>
   <span id="watching-behind" class="wbehind"></span>
   <span id="watching-what"></span>
-  <button id="watching-skip" class="ghost" type="button" title="Stop animating and jump the board to where the game actually is. Nothing is lost — every line is already in the History panel.">Skip</button>
 </div>
 
 <main>

src/web/splash.ts

@@ -40,6 +40,29 @@ if (heroImage && lightbox) {
  * other way costs a click and a lobby that says it cannot reach a server — which is legible, and
  * recoverable. So a slow or flaky probe leaves the door alone; only a definite answer closes it.
  */
+/**
+ * CARRY `?pace=` THROUGH THE DOORS — v0.8.0.3.
+ *
+ * Both doors are static hrefs that REPLACE the query string (`play.html?lobby`,
+ * `play.html?solitaire`), so a `pace` typed on this page was silently dropped on the way in: Jesse
+ * played a whole game believing he was at 7× when the play page had only ever seen `?lobby`. The
+ * durable answer is the speed control on the play screen, which persists per viewer — this keeps the
+ * URL lever honest for handing two playtesters different speeds, which is the only thing it was ever
+ * for.
+ */
+try {
+  const pace = new URLSearchParams(location.search).get('pace');
+  if (pace !== null) {
+    for (const door of Array.from(document.querySelectorAll('a.door'))) {
+      const href = door.getAttribute('href');
+      // Only the doors into the game, and only ones that have not been disabled above.
+      if (href?.startsWith('./play.html?')) door.setAttribute('href', `${href}&pace=${encodeURIComponent(pace)}`);
+    }
+  }
+} catch {
+  // A door that keeps its own href is the status quo, not a broken page.
+}
+
 const mpDoor = document.getElementById('door-multiplayer');
 if (mpDoor) {
   const close = (): void => {

src/web/step-queue.ts

@@ -19,7 +19,8 @@
 
 import type { PublicFrame } from '../sim/view.ts';
 import type { DisplayStep } from '../sim/display-step.ts';
-import { applyPublicDelta } from '../sim/public-delta.ts';
+import { applyPublicDelta, changedPiles } from '../sim/public-delta.ts';
+import type { PileKey } from '../sim/public-delta.ts';
 import { dwellForStep } from '../sim/pacing.ts';
 
 export type StepQueue = {
@@ -43,6 +44,13 @@ export type StepQueue = {
   behind(): number;
   /** The last step actually shown, for the caption line (#15). Null before anything has been shown. */
   showing(): DisplayStep | null;
+  /**
+   * The piles the step now on screen moved, for the display to light.
+   *
+   * Here because this is the only place that holds both the frame before a step and the frame after
+   * it — deriving it anywhere else would mean keeping a second copy of the board in step.
+   */
+  lit(): readonly PileKey[];
   /** True while there is anything left to show. */
   busy(): boolean;
 };
@@ -65,6 +73,7 @@ export function createStepQueue(
 ): StepQueue {
   let shown: PublicFrame | null = null;
   let last: DisplayStep | null = null;
+  let litPiles: readonly PileKey[] = [];
   let pending: DisplayStep[] = [];
   /** When the step now on screen is due to give way. Null when nothing is waiting. */
   let dueAt: number | null = null;
@@ -75,8 +84,14 @@ export function createStepQueue(
 
   /** Applies one step to the displayed board. A step's delta chains off the previous step's frame. */
   const show = (step: DisplayStep): void => {
+    const before = shown;
     shown = applyPublicDelta(shown, step.frame);
     last = step;
+    /**
+     * NOT FOR YOUR OWN MOVES. You drew that card; you do not need the deck flashed at you. Same rule
+     * that gives your own steps no dwell — the display is for watching everybody else.
+     */
+    litPiles = step.player !== null && step.player === viewer() ? [] : changedPiles(before, shown);
   };
 
   return {
@@ -84,6 +99,8 @@ export function createStepQueue(
       shown = frame;
       pending = [];
       dueAt = null;
+      // Nothing was watched arriving at this board, so nothing on it is lit.
+      litPiles = [];
       // `last` deliberately survives: a reconnect should not blank the caption line, and the
       // sentence describing the most recent action is still true.
     },
@@ -136,6 +153,7 @@ export function createStepQueue(
     current: () => shown,
     behind: () => pending.filter((s) => dwell(s) > 0).length,
     showing: () => last,
+    lit: () => litPiles,
     /**
      * STILL SHOWING SOMETHING, not just still holding something back.
      *

test/pacing.test.ts

@@ -14,7 +14,7 @@ import { readFileSync } from 'node:fs';
 import { dirname, join } from 'node:path';
 import { fileURLToPath } from 'node:url';
 
-import { DWELL, MAX_PACE, dwellFor, dwellForStep, kindOf, watchableCount } from '../src/sim/pacing.ts';
+import { DWELL, MAX_PACE, PACE_LEVELS, dwellFor, dwellForStep, kindOf, watchableCount } from '../src/sim/pacing.ts';
 import type { StepKind } from '../src/sim/pacing.ts';
 import type { Intent } from '../src/engine/intents.ts';
 
@@ -52,7 +52,7 @@ describe('pacing — dwell by kind', () => {
     assert.equal(kindOf('switch.end'), 'bookkeeping');
     /**
      * `localOps.choose` IS AN ANNOUNCEMENT, not bookkeeping — moved 2026-09-09 after the first real
-     * play on `phoenix.local`. It is the line reading "Player Bot 1 chose to SWITCH", the heading for
+     * play on the test server. It is the line reading "Player Bot 1 chose to SWITCH", the heading for
      * everything that follows, and at zero dwell a bot's turn began with no sign of what it was about
      * to do.
      */
@@ -120,6 +120,37 @@ describe('pacing — dwell by kind', () => {
     assert.equal(watchableCount(queue, 0), 0, 'with animation off, nothing is behind');
   });
 
+  it('offers speeds a player actually reached for, and none the code would clamp', () => {
+    /**
+     * Jesse played a whole game believing he was at 7× and was in fact at 1×: `?pace=` shipped as the
+     * only lever, and `index.html`'s doors are `play.html?lobby` / `play.html?solitaire`, so arriving
+     * from the splash REPLACES the query string. Hence a real control on the play screen, and hence
+     * this ladder — which must reach the speeds people ask for and must not offer one that
+     * `dwellFor` would silently clamp.
+     */
+    assert.equal(PACE_LEVELS[0], 0, 'off must be the first rung — #18 wants it turned off');
+    assert.ok(PACE_LEVELS.includes(1), 'the default must be on the ladder');
+    assert.ok(PACE_LEVELS.includes(7), '7x was asked for by name');
+    /**
+     * The ceiling is not theoretical. Jesse played at 10× — the top of the ladder as it then was —
+     * and called it "still a bit fast, but followable", so the ladder has to go past the speed
+     * somebody actually reached for and found insufficient.
+     */
+    assert.ok(PACE_LEVELS.some((p) => p > 10), 'the ladder must go beyond the speed that was too fast');
+    for (const p of PACE_LEVELS) {
+      assert.ok(p <= MAX_PACE, `${p}x is past MAX_PACE, so the control would lie about it`);
+      assert.equal(dwellFor('switch.move', p), Math.round(DWELL.switching * p));
+    }
+    // Strictly increasing, so stepping the control always changes the speed.
+    for (let i = 1; i < PACE_LEVELS.length; i++) {
+      assert.ok(PACE_LEVELS[i]! > PACE_LEVELS[i - 1]!, 'the ladder must be strictly increasing');
+    }
+    // The slowest rung has to be slow enough to be worth having: six switching moves at the top of
+    // the ladder is a full minute, which is the "watch them struggle" case.
+    const slowest = dwellFor('switch.move', PACE_LEVELS[PACE_LEVELS.length - 1]!) * 6;
+    assert.ok(slowest >= 60_000, `the slowest a switching turn can be watched is ${slowest}ms`);
+  });
+
   it('a silent step beats only when the clock turns over — TODO #18', () => {
     /**
      * Both obvious rules were wrong, so both are pinned. "No narration, no dwell" flashed past
@@ -127,18 +158,55 @@ describe('pacing — dwell by kind', () => {
      * that changed the board" beat on every turn hand-off — `submit()` steps `advance()` about 4.6
      * times per intent — which came to a quarter of an hour a game.
      */
-    const silent = { cause: 'phase' as const, lines: [] as string[] };
+    const silent = { cause: 'phase' as const, player: null, lines: [] as string[] };
     assert.equal(dwellForStep({ ...silent, frame: { table: { actor: 2 } } }), 0, 'a turn hand-off shows nothing');
     assert.equal(dwellForStep({ ...silent, frame: { table: {} } }), 0, 'a step that changed nothing shows nothing');
     assert.equal(dwellForStep({ ...silent, frame: { table: { phase: 'mainline' } } }), DWELL.phase);
     assert.equal(dwellForStep({ ...silent, frame: { table: { stage: 4 } } }), DWELL.phase);
     // Narration always earns the dwell of whatever caused it, clock or no clock.
     assert.equal(
-      dwellForStep({ cause: 'switch.move', lines: ['moved'], frame: { table: {} } }),
+      dwellForStep({ cause: 'switch.move', player: 1, lines: ['moved'], frame: { table: {} } }),
       DWELL.switching,
     );
   });
 
+  it('the speed control stretches the clock at a THIRD of the rate it stretches people', () => {
+    /**
+     * Two complaints, one from each direction, and the answer is between them.
+     *
+     * v0.8.0.3, from a 5× game: *"after my turn … I'm still subject to that same delay before it
+     * moves on. That makes no sense."* — phases were scaling with everything else and walling off a
+     * player's own turn. So they were pinned at their tabled beat.
+     *
+     * v0.8.0.7, from a 10× game: *"phases displayed on the upper line go by too quickly still.
+     * Should be 4 times as long — at a guess."* — pinned was too short to read the caption.
+     *
+     * Damped scaling satisfies both: 1× unchanged, 10× lands exactly on the four-times guess, and
+     * the cost stays bounded because phase beats cluster rather than accumulate.
+     */
+    const phase = { cause: 'phase' as const, player: null, lines: ['New Train'], frame: { table: { phase: 'newTrain' } } };
+    const theirs = { cause: 'switch.move' as const, player: 1, lines: ['moved'], frame: { table: {} } };
+
+    assert.equal(dwellForStep(phase, 1), DWELL.phase, '1x must be exactly the tabled beat');
+    assert.equal(dwellForStep(phase, 10), DWELL.phase * 4, '10x must be four times it, as asked for');
+
+    for (const pace of [2, 3, 5, 7, 10, 15, 20]) {
+      const p = dwellForStep(phase, pace);
+      const t = dwellForStep(theirs, pace);
+      assert.ok(p > DWELL.phase, `a phase must grow at ${pace}x`);
+      assert.ok(
+        p < DWELL.phase * pace,
+        `a phase must grow SLOWER than the multiplier at ${pace}x, or the clock walls off the turn`,
+      );
+      assert.ok(t > p, `somebody's move must still outlast a phase beat at ${pace}x`);
+    }
+    // Off still means off, for the clock as much as for anybody's move; and below 1x the clock
+    // follows the multiplier straight, because "faster" should mean everything.
+    assert.equal(dwellForStep(phase, 0), 0);
+    assert.equal(dwellForStep(theirs, 0), 0);
+    assert.equal(dwellForStep(phase, 0.5), DWELL.phase * 0.5);
+  });
+
   it('a real switching turn is watchable in a few seconds, not tens of them', () => {
     // Six moves is the engine's cap per crew ("N of 6 Moves left"), so this is the worst ordinary
     // case for one crew: the announcement, six moves, and an end that shows nothing.
@@ -155,7 +223,7 @@ describe('pacing — dwell by kind', () => {
 
   it("a bot's ordinary turn is followable, which is what the first real play was not", () => {
     /**
-     * MEASURED FROM A REAL GAME, then pinned. Jesse, after installing v0.8.0 on `phoenix.local`:
+     * MEASURED FROM A REAL GAME, then pinned. Jesse, after installing v0.8.0 on the test server:
      * *"bot play was way too fast. I briefly saw that it was the bot's office area then their turn
      * was done."* This is the shape that turn actually had — no switching in it at all, because
      * switching is not legal until there is track down — and under the original values it came to

test/public-delta.test.ts

@@ -17,7 +17,7 @@ import type { GameConfig, GameState, PlayerIndex } from '../src/engine/state.ts'
 import { applyIntent } from '../src/engine/apply.ts';
 import { currentActorOfState, publicSnapshot } from '../src/sim/view.ts';
 import type { PublicFrame } from '../src/sim/view.ts';
-import { applyPublicDelta, deltaPublicFrame } from '../src/sim/public-delta.ts';
+import { applyPublicDelta, changedPiles, deltaPublicFrame } from '../src/sim/public-delta.ts';
 
 const config: GameConfig = {
   mode: 'competitive',
@@ -184,3 +184,73 @@ describe('public frame delta', () => {
 });
 
 type PublicIndexed = { seat: number; before: PublicFrame; after: PublicFrame };
+
+describe('which piles a step moved', () => {
+  /**
+   * MEASURED FROM REAL PLAY, then pinned. The table in `changedPiles` claims what each action moves,
+   * and a claim in a comment is worth nothing unless something checks it — so this drives real games
+   * and asserts the mapping holds, action by action.
+   */
+  it('maps each action to the piles it actually touches', () => {
+    const seen = new Map<string, Set<string>>();
+    /**
+     * TWO PASSES, because a single driver cannot reach every case. Left to itself the bot almost
+     * never takes a Department card, and a driver that prefers one then never draws from the deck —
+     * so each preference is played out separately and the assertions below require BOTH to have
+     * been observed rather than passing on whichever happened to occur.
+     */
+    for (const prefer of ['draw.fromDepartment', 'draw.fromHomeOffice'] as const) {
+      for (const seed of [1917398, 191056, 4242]) {
+        const s = newState(seed);
+        for (let i = 0; i < 400; i++) {
+          const actor = currentActorOfState(s);
+          if (actor === null) break;
+          const options = legalActions(s, actor);
+          if (options.length === 0) break;
+          const chosen =
+            options.find((o) => o.type === prefer) ??
+            options.find((o) => o.type === 'card.discard') ??
+            options[i % options.length]!;
+          const before = publicSnapshot(s);
+          const r = applyIntent(s, actor, chosen);
+          if (!r.ok) break;
+          pump(s);
+          const piles = changedPiles(before, publicSnapshot(s)).map((p) => p.replace(/dept\d/, 'dept'));
+          if (!seen.has(chosen.type)) seen.set(chosen.type, new Set());
+          for (const p of piles) seen.get(chosen.type)!.add(p);
+          if (piles.length === 0) seen.get(chosen.type)!.add('(none)');
+        }
+      }
+    }
+
+    const of = (t: string): Set<string> => seen.get(t) ?? new Set();
+    // NOT VACUOUS: the four cases the mapping is actually about must all have happened.
+    for (const needed of ['draw.fromHomeOffice', 'draw.fromDepartment', 'card.discard', 'card.play']) {
+      assert.ok(of(needed).size > 0, `${needed} never occurred, so its rule proved nothing`);
+    }
+
+    // A HOME OFFICE DRAW MOVES THE COUNT AND NOTHING ELSE ON A PILE. The card is private; the deck
+    // getting shorter is not, and it is the only thing a watcher can be shown.
+    assert.deepEqual([...of('draw.fromHomeOffice')].sort(), ['home']);
+    // A DEPARTMENT DRAW touches that Department, and sometimes the deck too — the pile refills from
+    // it. Both are public, so both may light.
+    for (const p of of('draw.fromDepartment')) {
+      assert.ok(p === 'dept' || p === 'home', `a Department draw moved "${p}"`);
+    }
+    assert.ok(of('draw.fromDepartment').has('dept'), 'a Department draw must light its Department');
+    // A DISCARD lands face up on a Department, and which one is public.
+    assert.deepEqual([...of('card.discard')].sort(), ['dept']);
+    // A PLAYED CARD that does not stay on the board lands face up in the Salvage Yard.
+    assert.ok(of('card.play').has('salvage'), 'a played card must be able to light the Salvage Yard');
+    // ENDING A PHASE moves no card anywhere, so nothing should light for it.
+    for (const quiet of ['draw.end', 'loadUnload.end', 'switch.end', 'localOps.choose']) {
+      if (of(quiet).size > 0) assert.deepEqual([...of(quiet)], ['(none)'], `${quiet} lit a pile`);
+    }
+  });
+
+  it('lights nothing without a previous frame to compare against', () => {
+    // A reset has nothing to have watched arriving, so nothing on it is lit.
+    const s = newState(4242);
+    assert.deepEqual(changedPiles(null, publicSnapshot(s)), []);
+  });
+});

test/redaction.test.ts

@@ -524,6 +524,9 @@ describe('#91 — nothing private survives serialisation, in any state', () => {
       // The rules the game was dealt under, and the score.
       'houseRules', 'mode', 'optionalRules', 'days', 'minCombinedRevenue',
       'maxCollisionsPerDay', 'maxCollisionsTotal', 'collisionsToday', 'collisionsTotal',
+      // What the Day that just ended finished on. Public for the same reason the running counts are:
+      // a collision happens on the Mainline in front of everybody.
+      'collisionsPrevDay',
       'status', 'outcome', 'extraDays', 'extensionVotes', 'official', 'tally',
       // Names, seats, revenue and HAND SIZE — never hand contents.
       'players',

test/step-queue.test.ts

@@ -179,7 +179,7 @@ describe('the step queue', () => {
      * REGRESSION. `busy()` was `pending.length > 0`, so the instant the final step of a burst was
      * shown the queue reported idle: the animation loop stopped and the district panel snapped back
      * to the viewer's own board without that step ever being looked at. Jesse, from the first real
-     * play on `phoenix.local`: *"I briefly saw that it was the bot's office area then their turn was
+     * play on the test server: *"I briefly saw that it was the bot's office area then their turn was
      * done and it pointed back to my office area"*, and the countdown row appeared "very briefly".
      *
      * The panel follows `busy()`, so this is the property that keeps somebody else's board on screen
@@ -251,6 +251,25 @@ describe('the step queue', () => {
     assert.ok(q.showing() !== null, 'the caption should survive a reset');
   });
 
+  it('can always be emptied, so a player is never stranded behind it', () => {
+    /**
+     * "Your Move" is put away while the board is catching up (v0.8.0.6), which makes `busy()` the
+     * thing standing between a player and their own turn. So the ways it can be cleared matter more
+     * than they did: `skip()` must always work, from any state, including one where the clock has
+     * never advanced — which is exactly the situation a page with no `requestAnimationFrame` is in,
+     * and how this was found.
+     */
+    const { steps, final } = realSteps(1917398, 400);
+    const q = createStepQueue();
+    q.reset(baseline(1917398));
+    q.push(steps);
+    // Never advanced at all: no frame has been shown, and the queue is full.
+    assert.equal(q.busy(), true);
+    assert.equal(q.skip(), true, 'a never-advanced queue must still be skippable');
+    assert.equal(q.busy(), false, 'and must be idle afterwards, or the player stays locked out');
+    assert.deepEqual(q.current(), final);
+  });
+
   it('draws nothing before a reset has arrived', () => {
     const q = createStepQueue();
     assert.equal(q.current(), null);

test/web.test.ts

@@ -23,7 +23,7 @@ import { variantsFor } from '../src/engine/track.ts';
 import { divisionSvg, officeSvg } from '../src/sim/board-svg.ts';
 import type { DivisionView } from '../src/sim/view.ts';
 import { ENHANCEMENT_RULES, STAGES_PER_DAY } from '../src/engine/content.ts';
-import { dayEndHtml, facilitiesHtml, resultsHtml, timetableHtml } from '../src/web/panels.ts';
+import { dayEndHtml, facilitiesHtml, pilesHtml, resultsHtml, timetableHtml } from '../src/web/panels.ts';
 import { turnChartHtml } from '../src/sim/turnchart.ts';
 import { fieldSelectors } from '../src/web/settings-form.ts';
 import { record, renderHtml } from '../src/sim/replay.ts';
@@ -3614,6 +3614,136 @@ describe('the Day rolling over says so (Gitea#10)', () => {
     assert.ok(html.includes('3 Days left'), `the Days remaining are wrong:\n${html}`);
   });
 
+  it('draws the Home Office deck, face down, with its count', () => {
+    /**
+     * `f.deck` has carried the face-down count since the Frame existed and NOTHING drew it — the
+     * exact display gap `test/display-gaps.test.ts` sweeps for, surviving in the panel that draws
+     * every other pile. Asked for by Jesse 2026-09-10, who also wanted somewhere for a draw to
+     * flash: taking a card off this deck is the commonest move nobody can see.
+     */
+    const s = createEngineGame({
+      id: 'piles',
+      seed: 5,
+      config: {
+        mode: 'competitive', days: 5, minCombinedRevenue: 60,
+        maxCollisionsPerDay: 0, maxCollisionsTotal: 0, pvpCardsAllowed: false,
+        optionalRules: { reducedVisibility: false, employeeRotation: false, emergencyToolbox: false },
+      },
+      playerNames: ['Joe', 'Bot 1'],
+    });
+    const f = snapshot(s, [], null);
+    assert.ok(f.deck > 0, 'the deal should leave cards in the Home Office deck');
+
+    const html = pilesHtml(f);
+    assert.ok(html.includes('Home Office'), `no Home Office pile:\n${html}`);
+    assert.ok(html.includes(`>${f.deck}<`), 'the face-down count is not shown');
+    // Face down means the card slot must NOT name a card — that is the whole point of the pile.
+    assert.ok(html.includes('facedown'), 'the Home Office pile is not marked face down');
+    assert.ok(html.includes('face down'), 'the card slot should say so rather than naming a card');
+    // It comes first: a card travels out of here, then onto a Department or the Salvage Yard.
+    assert.ok(
+      html.indexOf('Home Office') < html.indexOf('Dept 1'),
+      'the draw deck should be read before the piles cards land on',
+    );
+  });
+
+  it('names the train on top of the Salvage Yard, instead of "a card"', () => {
+    /**
+     * The Salvage Yard is a FACE-UP pile and its tile reads the top card — but `trainScheduled`
+     * pushes a synthetic `train-<number>` id rather than the id of the card that was played
+     * (`apply.ts`), so there was nothing in `s.cards` to look up and the tile said "a card". A train
+     * is scheduled several times a Day, so that id is on top most of the time: the pile was face up
+     * and had nothing to say. Jesse, 2026-09-10: *"why is salvage deck not face up. I should see the
+     * card played onto salvage."*
+     */
+    const s = createEngineGame({
+      id: 'salv',
+      seed: 5,
+      config: {
+        mode: 'competitive', days: 5, minCombinedRevenue: 60,
+        maxCollisionsPerDay: 0, maxCollisionsTotal: 0, pvpCardsAllowed: false,
+        optionalRules: { reducedVisibility: false, employeeRotation: false, emergencyToolbox: false },
+      },
+      playerNames: ['Joe', 'Bot 1'],
+    });
+    s.decks.salvageYard.push('train-13');
+    const f = snapshot(s, [], null);
+    assert.equal(f.salvage.top, 'Train 13', 'a scheduled train on the pile must be named');
+
+    const html = pilesHtml(f);
+    assert.ok(html.includes('Train 13'), `the Salvage tile does not name the train:\n${html}`);
+    assert.equal(html.includes('>a card<'), false, 'the face-up pile still says "a card"');
+  });
+
+  it('lights only the pile a watched move touched', () => {
+    const s = createEngineGame({
+      id: 'piles2',
+      seed: 5,
+      config: {
+        mode: 'competitive', days: 5, minCombinedRevenue: 60,
+        maxCollisionsPerDay: 0, maxCollisionsTotal: 0, pvpCardsAllowed: false,
+        optionalRules: { reducedVisibility: false, employeeRotation: false, emergencyToolbox: false },
+      },
+      playerNames: ['Joe', 'Bot 1'],
+    });
+    const f = snapshot(s, [], null);
+
+    assert.equal(pilesHtml(f).includes('pilelit'), false, 'nothing is lit when nothing was watched');
+
+    const home = pilesHtml(f, ['home']);
+    assert.equal((home.match(/pilelit/g) ?? []).length, 1, 'exactly one pile should light');
+    assert.ok(
+      home.indexOf('pilelit') < home.indexOf('Dept 1'),
+      'a Home Office draw must light the Home Office pile, not a Department',
+    );
+
+    const dept2 = pilesHtml(f, ['dept1']);
+    assert.equal((dept2.match(/pilelit/g) ?? []).length, 1);
+    assert.ok(dept2.indexOf('Dept 2') > dept2.indexOf('Dept 1'), 'order sanity');
+    // Two piles can move at once — a Department draw that refills from the deck.
+    assert.equal((pilesHtml(f, ['home', 'dept0']).match(/pilelit/g) ?? []).length, 2);
+  });
+
+  it('reports the ENDED Day\'s collisions, not the fresh Day\'s zero', () => {
+    /**
+     * Jesse, 2026-09-09, at the end of a Day 1 with two collisions in it: *"It shows a total of two
+     * collisions, but zero today. Since we just finished day one, that does seem to be a
+     * contradiction."*
+     *
+     * The cause is a one-line ordering fact: `advance.ts` increments the Day and then zeroes
+     * `collisionsToday`, and this dialog is drawn from the frame whose Day went UP — so it read the
+     * fresh Day's zero and printed it beside a running total that could not agree with it. The count
+     * is captured at the rollover now, and the dialog names the Day rather than saying "today".
+     */
+    const s = createEngineGame({
+      id: 'collide',
+      seed: 5,
+      config: {
+        mode: 'competitive',
+        days: 5,
+        minCombinedRevenue: 60,
+        maxCollisionsPerDay: 3,
+        maxCollisionsTotal: 10,
+        pvpCardsAllowed: false,
+        optionalRules: { reducedVisibility: false, employeeRotation: false, emergencyToolbox: false },
+      },
+      playerNames: ['Joe', 'Bot 1'],
+    });
+    // The state as the rollover out of Day 1 leaves it: two collisions happened, `today` is reset.
+    s.clock.day = 2;
+    s.collisionsPrevDay = 2;
+    s.collisionsToday = 0;
+    s.collisionsTotal = 2;
+
+    const html = dayEndHtml(snapshot(s, [], null));
+    assert.ok(html.includes('Day 1 has ended'), `wrong Day named:\n${html}`);
+    assert.ok(html.includes('<b>2</b> on Day 1'), `the ended Day's collisions are wrong:\n${html}`);
+    assert.ok(html.includes('<b>2</b> in all'), `the running total is wrong:\n${html}`);
+    assert.doesNotMatch(html, /<b>0<\/b> today/, `still reporting the fresh Day's zero:\n${html}`);
+    // The contradiction itself: a Day-end dialog must never claim fewer in all than on that Day.
+    assert.doesNotMatch(html, /<b>0<\/b> on Day 1/, 'reported no collisions on a Day that had two');
+  });
+
   it('counts the last Day as the last Day rather than promising more', () => {
     const html = dayEndHtml(frameAt(6));
     assert.ok(html.includes('Day 5 has ended'), 'the final Day is misnamed');