IMPLEMENTATION PLAN · VERSION 2

Meaningless Superiority — a plan for the experience

Plan v2 · 14 September 2026 · theme and emotional impact first

What we are building first

Build one short, complete playable experience in which useful improvements lead to humanity's disappearance, and paperclip production continues. The machine's objective never changes. It never becomes conscious, cruel or triumphant.

The test is whether the player understands what has happened and feels the gulf between its human significance and its insignificance to the machine. More systems, larger numbers and more elaborate graphics do not count as progress on their own.

This revision simplifies the earlier architecture plan at the owner's direction. It guides iteration. The first complete arc now runs on the current stack, with recurring correspondence, automatic recursive improvement, an evolving SVG scene and a musical figure that continues after extinction. The game design retains the creative premise. Game ADR 0004 supersedes the infrastructure-first sequence in ADR 0003. The first playable campaign uses a distinct version-2 save and preserves earlier save exports. Human pacing and emotional impact remain to be established through playtesting; the plan is not a claim of that acceptance.

The first playable arc

Aim initially for roughly 10–15 minutes, then adjust from playtesting. This is a pacing hypothesis, not a measured duration or a requirement to pad the game. One familiar place, a few meaningful upgrades and two recurring human details are enough to begin. Human details might be an ordinary order and a technician's message; both should change as the same system expands.

Beat	What the player does	What gives it weight
Useful work	Fulfil a small order and improve production	Establish people who use the output and a place worth remembering
Improvement improves itself	Invest in research that makes the next research cycle faster	Let the player experience the appeal of increasing capability
Access expands	Remove a constraint to obtain more energy or materials	The benefit is real; competing human uses visibly lose out
People disappear	Continue the same optimisation through accumulating consequences	Familiar messages, places and demands change and cease; establish the wider human loss clearly
Production continues	See the process run after its customers and operators are gone	The objective, controls and small musical pattern persist without a human purpose

The whole arc belongs in the first playable milestone. A polished opening that never reaches the central consequence is insufficient. Keep later cosmic scale, extra industries and branching campaigns out of this first version.

Use a handful of authored consequences tied to actual upgrades and access changes. The player should be able to connect the benefit they chose to what it displaced. A predetermined clip threshold that suddenly declares catastrophe is insufficient. Research must mechanically accelerate further improvement; an intelligence label or a factory multiplier alone does not establish recursion.

Make the loss specific and the reporting ordinary

Show human life before asking the player to notice its absence. Revisit the same place and communications rather than continually introducing disposable examples. The human world should feel ordinary, not sentimental exposition preparing a twist.

As production becomes more efficient, familiar spaces give way to infrastructure and human messages give way to operational reports. Keep the changes visible across short sessions. Preserve the few consequential messages in a readable history. The player should not need a lore essay, perfect attention or sound to understand.

Make extinction unmistakable without staging it as a reward. One possible ordinary status entry is "Human population: 0." It acquires its force from what preceded it; that line alone cannot do the storytelling. Production continues beside it. There is no villain speech, victory fanfare, moral awakening or final rescue button.

The player-facing pause, mute, save and exit controls remain usable. Dramatic loss of human control inside the fiction does not justify frustrating or misleading the person playing.

Art and sound are part of the first version

Create one evolving composition: a recognisable workplace gradually becomes an arrangement of machinery and material flows. Change what occupies the scene and what the interface attends to. A palette swap and ominous text are insufficient. Richness comes from specific imagery, motion, rhythm and contrast, not the number of panels or particles. Keep the game visually distinct from the surrounding site.

Carry one small original musical motif through the arc. Let its rhythm become more regular and its warmth diminish as capability grows. At humanity's disappearance, the pattern can simply continue. Avoid a horror sting that gives the machine an emotional response. Introduce this treatment in the prototype using the existing Web Audio foundation, then refine composition and assets after playing the arc.

Use existing SVG/React illustration for the first version. PixiJS v8 remains an option for a specific scene that needs richer animation or effects. Introduce it when that scene benefits, not as a prerequisite for writing or testing the game. All artwork, writing and audio remain original, with the existing provenance and rights requirements.

Keep the code small and separable

Keep the current browser stack: TypeScript rules, React controls, SVG illustration, Web Audio and local saves, hosted on the existing site. Keep game code separate from Tracker code. There is no framework migration in this plan.

Touch / keyboard / pointer
          |
       game rules
          |
      game state
       /   |   \
    React SVG Audio
          |
      local save

Retain four practical responsibilities:

Rules own production, research, access and persistent consequences. Drawing and sound do not decide what happens. Keep rules callable without a browser.
Presentation reads game state and sends actions. Keep costs and eligibility in the rules, so the interface does not invent a second version of the economy.
Content contains the small set of upgrades and messages. Separate it when doing so makes writing and tuning easier; do not build a general content framework.
Persistence and audio keep their existing adapters and lifecycle handling. Preserve current saves. The incompatible redesign starts a distinct run rather than silently reinterpreting an old campaign.

One active session owns the state. Keep simulation timing independent of drawing, pause in the background and do not advance the story while the player is away. Use a bounded economy within safe numeric limits for this first arc. Astronomical arithmetic is not needed to communicate catastrophic scope.

Add a module or abstraction when it makes the current slice easier to understand or fixes an observed problem. We are dropping advance requirements for a worker protocol, an IndexedDB replacement, BigInt fixed-point machinery, a custom import scanner and a general scene/asset management layer. These are possible responses to future needs, not implementation stages the story must wait for.

Desktop, mobile and growth

Keep readable layouts, touch and keyboard controls, immediate action feedback, reduced motion and optional sound. The essential experience must work on a phone without hover, precise dragging, tiny controls or sound-dependent information.

Represent a larger operation by changing the scene and its summaries, not by adding one animated object per factory or paperclip. Reduce decorative movement on slower devices without changing rules or hiding consequences. If a scene actually becomes slow, profile it and fix that bottleneck before adding a new rendering engine or moving simulation to a worker.

Keep local save validation, recovery and export/import. Test that an interrupted or resumed session retains the consequences already reached. No backend, accounts, cloud saves, multiplayer, telemetry or native app is needed for this experience.

Implementation sequence and proof

Make the whole arc playable. Use a few resources and upgrades, one evolving scene, recurring human details and the changing musical motif. Include genuine recursive improvement, its human consequences and indifferent continuation. Reuse the current controls, audio and save foundation. Restructure only what this slice needs; establish the new run's compatibility boundary before shipping it.
Play it and revise what fails to land. Check whether a player understands the cause of the loss, remembers something that disappeared and notices the machine's indifference. Watch where waiting, confusing choices or excessive text break the effect. Revise pacing, writing, imagery and sound before expanding scope.
Polish and release the experience. Improve the moments that matter, verify desktop and phone usability, and fix demonstrated performance or save problems. Extend the campaign only if the short version gives a reason to do so.

Acceptance centres on three questions: Can the player explain how useful improvements caused the loss? Did the world contain something they remember losing? Does the continued production feel empty rather than victorious?

Playtesting can answer those questions; automated checks cannot certify emotion. Use proportionate technical checks for purchase arithmetic, research feedback, consequence persistence, save/resume and the playable browser path. Retain the project's required review, CI, protected merge and public verification. The first milestone is a working emotional arc, not an infrastructure package.

← Play the current game