JesseMarkowitzandClaude Opus 5
5c1784d990
A 182-turn session spent 30% of its turns below zero, went negative on turn 18 and never recovered, and chose the same hard-times option — "take every shift going" — thirty-seven times. The cause was a tuning change made without re-deriving what it meant. Rent went from 385 to 455 while chasing a different problem, which took the mailroom baseline from the designed -$3/day to -$13/day. Worse, at the debt that run accumulated the weekly service charge came to $180 against an escape option worth $120. The hole was inescapable by arithmetic, whatever the player did. Nothing caught it. Every archetype either optimised money or got promoted out of the problem before it bit, and the dominance test passed because rent_bounced was only 17% of turns. Fixes: mailroom wages 980 -> 1120, restoring the -$3/day baseline; the escape option 120 -> 220, so it is worth more than a week's rent; rent_bounced from weight 5000 with no cooldown to 900 on a cooldown of 3, because being broke should colour a run rather than replace it; debt_collector from weight 25 to 45, since at 25 it appeared six times in 262 turns and debt became a ratchet. Three guards so this class of bug cannot recur quietly. The baseline is now asserted directly: a test sums the pack's own upkeep over twenty fortnights and requires the mailroom to net between -90 and +10, and Dispatch to be better but not so much better that money stops mattering. That would have failed the moment rent changed. Outcome tests over simulated play are a slow and noisy way to detect a number that is simply wrong. A `lifer` archetype refuses any option that would change stage — generically, by looking for an effect on `stage` — and otherwise plays for people. It reproduces the session that found this; no other archetype can. And no archetype may spend more than a quarter of a run below zero. Hard times is a state a player passes through; living in it is the failure mode. Also fixed: the analyser reported "every ~-5 turns". One export can hold several playthroughs and turn numbers restart with each, so spans are now accumulated per run and pooled rather than measured across the seam between two. 121 tests. Reasoning in docs/DECISIONS.md §29-30. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01VMSFHyVPitUoosW5wyEADj
7.5 KiB
The Ladder
A turn-based career simulation game that runs entirely in the browser. You start at the bottom of an organisation and play turns — a day at a time — making choices that move your money, your skills and your standing with the people around you. There is no win state and no game over: a bad run means worse options, not an ending, and you can stop whenever you like and read the retrospective.
The engine knows nothing about offices, mailrooms or bosses. A setting is a
content pack — characters, events, choices, stat names and tuning, all data.
corporateladder is the default pack; a Wild West or lemonade-stand pack would
be new data and no new code.
Status
Engine and the first content pack are built and tested. No UI yet — the game is currently playable only from a test harness.
| Piece | State |
|---|---|
| Path-addressed state, conditions, effects | done |
| Seeded RNG, save/resume | done |
| Event selection, option gating | done |
| Turn loop, per-turn upkeep | done |
| Content validator + reachability tests | done |
| Corporate Ladder pack — 31 events, two tiers | done |
| Turn screen, retrospective, feedback widget | done |
| Local-storage save, file export/import | done |
| Single-file build | done |
| Calendar, scheduled wages and rent | done |
| Promotion from the mailroom to Dispatch | done |
Repetition analytics (npm run analyze) |
done |
| Version and build shown on screen | done |
| Keyboard play (Enter / Tab) | done |
| Played in a real browser | done — first playtest 2026-09-09 |
The first playtest confirmed the interface, saves and the exported log all work in a real browser. Its findings — an unexplained daily drain, no sense of the week, and an economy nobody could get ahead in — are what the calendar and the retuned economy are for.
Note that the development machine has no browser that can render a page, so the UI tests drive the real modules against a minimal fake DOM. That covers wiring, not layout: styling and dark mode are only ever verified by a human opening it.
Playing it
npm run serve # then open http://localhost:8080
The game plays from the keyboard: Enter takes the focused option, Enter again moves to the next day, and Tab / Shift+Tab pick a different option. The top of every screen names the game, the pack, your current title, and the version and commit you are running.
A server is needed during development because browsers refuse ES module imports
over file://. If port 8080 is taken, run python3 -m http.server <port>
instead.
To hand the game to someone else:
npm run build # writes dist/theladder.html
That is one self-contained file with the stylesheet and every module inlined. It plays by double-clicking it — no server, no network, nothing installed.
npm test # 121 tests: engine, content, UI wiring, and the build
To see what a playtest actually did — turns against distinct situations, how often each recurred, which locked gates players kept meeting, and anything they typed into the feedback box:
npm run analyze # the newest export in logs/
npm run analyze -- <file> # a specific one
Drop exports into logs/. They are gitignored — a log is a record of what a
real person did, including anything they typed into the feedback box.
There are no runtime dependencies and no build step for development. Node is used only to run the tests, the single-file build, and the log analyser.
Layout
src/engine/ the game engine — imports nothing from content/, or from src/ui/
paths.js dotted-path get/set over state
conditions.js generic predicates: { path, op, value }, all/any/not
effects.js generic mutations: add/set/push/remove, with clamping
state.js pack -> schema and initial state
select.js which event fires this turn, which options are available
game.js the turn loop
rng.js seeded, serializable RNG
validate.js load-time content checking
src/ui/ screens, formatting and the app controller
app.js the only module that names a content pack
screens/ the turn screen and the retrospective
format.js state paths and conditions rendered as English
src/io/ local-storage save, file export/import, the feedback log
content/ content packs (settings)
tools/build.js flattens everything into one self-contained HTML file
test/ engine tests against a fixture pack; content and UI tests
docs/ DECISIONS.md — why things are the way they are
Planning/ the original design documents
The rule that keeps this honest: nothing in src/engine/ may import from
content/. The engine names no stat, no character and no setting of its own.
How content works
A pack declares what state exists and what it starts at, the relationship stats every character carries, the characters, and the events. Everything else falls out of that.
export default {
id: 'corporateladder',
startingStage: 'mailroom',
state: {
'money': { initial: 1200, min: -5000, max: 1000000 },
'skills.negotiation': { initial: 1, min: 0, max: 100 },
'flags.took_loan': { initial: false },
},
// Expanded per character into relationships.<id>.<stat>, so adding a
// character adds its relationship state automatically.
relationshipStats: {
likes_you: { initial: 50, min: 0, max: 100 },
fears_you: { initial: 50, min: 0, max: 100 },
wants_to_help_you: { initial: 50, min: 0, max: 100 },
},
characters: [
{ id: 'boss', name: '...', role_type: 'boss', description: '...' },
],
events: [
{
id: 'mail_run',
stage: 'mailroom',
weight: 100, // relative likelihood among eligible events
description: 'A cart of mail and two hours to move it.',
options: [
{
id: 'hustle',
label: 'Get it done fast.',
requires: [{ path: 'skills.negotiation', op: '>=', value: 3 }],
effects: [
{ path: 'money', op: 'add', value: 40 },
{ path: 'relationships.boss.likes_you', op: 'add', value: 3 },
],
},
],
},
],
};
An event may also carry once: true, a cooldown in turns, and requires
conditions of its own. There is no separate notion of a random event: every
turn draws from the whole eligible pool by weight.
Conditions may read engine bookkeeping (turn, stage, seen.<eventId>,
lastSeen.<eventId>) as well as pack state. Effects may not write it.
A pack also declares upkeep — a list of { label, requires?, effects }
applied at the end of a turn, after the choice. Adding every and offset
makes an entry periodic rather than daily: Corporate Ladder pays wages on
alternate Fridays ({ every: 14, offset: 4 }), charges rent every Monday, and
adds a loan service charge that steps up as the debt grows.
A pack may also declare a calendar:
calendar: {
cycleLength: 7,
cycleName: 'Week',
unitNames: ['Monday', 'Tuesday', 'Wednesday', 'Thursday', 'Friday', 'Saturday', 'Sunday'],
},
The engine derives calendar.index, calendar.cycle and calendar.name from
the turn number before each event is drawn, so content can require a Friday and
the interface can title the turn. Declare no calendar and there is no week.
Run validatePack(pack) when loading in development — it catches undeclared
paths, unknown characters, unusable ids, unreachable events and stages with no
fallback event, all of which otherwise fail silently mid-playthrough.