[ishaansathaye]

Make the visitor's input matter

doke's proposed timing experiment offered an explanation to operate, but little to create or change. This research PR explores experiences where the visitor makes something recognizable and the system responds: an inhabited world, a retained mark, or a journey through the body.

It contains eight concepts, three original playable mocks, source research, and a proposed Rust implementation path after a direction is selected. The application is unchanged. This branch starts directly from the original main baseline and does not depend on #1 or #2. Those PRs remain unmerged. This is a design review, not approval to implement or merge a direction.

Start here

• Research, sources, tradeoffs and next-phase scope
• Local gallery while the preview server is running: http://127.0.0.1:4180/
• To run elsewhere, check out this branch and run:

python3 -m http.server 4180 --bind 127.0.0.1 --directory docs/play/mocks

Open the local address in a browser. The mocks also open directly as HTML files. No build, account, data download or API key is needed. The localhost link is a local preview, not a public deployment.

Three experiences to try

Mock	Try this	Actual behavior and boundary
After a Bite	Drop bread or coffee into the body; compose a sequence, send a gut ripple, zoom into the gut, then replay.	Toy contents transport responds to local impulses. Taste, gut-to-brain and blood routes have distinct visual cues but authored pacing. No meal prediction, receptor-binding simulation or whole-body physiology.
Little Stranger	Paint food, draw a barrier in its path, then erase/repaint part of its world. Watch the transparent body and trail.	Sensors drive steering; contacts increase a bounded avoidance gain. Food is actually consumed. It is a fictional reactive controller, not a learned connectome or biological animal.
Memory Loom	Draw your own mark, keep it, scatter/rub away a piece and recall from the fragment. Hold several marks and try to confuse it.	Masked nearest-template retrieval can refuse weak, empty or ambiguous cues. It is not a neural network or human-memory simulation.

After a Bite

Little Stranger

Memory Loom

These are captures of working mocks, not generated concept images. Only three compact previews are committed; raw captures and test output are ignored. The sketches use Canvas/SDR glow; the existing Rust/Wasm/HDR application is unchanged.

Other ideas evaluated

The research also scopes Echo (local sound becomes a recurring acoustic sculpture), Your Tempo (a personal rhythm leaves a persistent signature), Borrowed Eyes (a virtual gaze explores an adapting field made from your image), A Day in Tides (composed events shift generic body rhythms), and Project Memory (sanitized change events organize a memory surface). Each has a reason to return and a specific failure risk.

Primary inspiration includes Phloem, NeuroMechFly, Growing Neural Cellular Automata, Particle Lenia, Lenia, Sandspiel, Simbrain, Karl Sims, Evolution, Chrome Music Lab and Dear Data. The research separates design inferences from those projects' findings. No external code, art, audio or trained weights were copied.

The food/body work draws on primary studies and NIDDK/NIDCD material, distinguishing human evidence from mouse/preparation results. It separates sensory information, local gut control, blood-borne substances and endocrine mechanisms. Food names do not establish personal neurotransmitter levels or a digital twin.

Recommendation and decision

My recommendation is Little Stranger for the strongest game, Memory Loom for the smallest personal-input experiment, and After a Bite for the closest extension of the brain–body vision. These are judgments to test, not demonstrated enjoyment. A body playground should first earn its scale through one satisfying local mechanic.

The next production PR should implement only the interaction the project owner wants to keep playing. The document proposes explicit playtest gates and a bounded, reproducible Rust model, with actual model signals driving the existing renderer. Accounts and leaderboards are deferred.

Review and verification

• Three agents contributed interaction research, neuroscience research and implementation/feasibility work. Separate implementation and science reviews were completed; their actionable findings were fixed.
• node tests/play-mocks.mjs passed locally in Chrome: transport/replayed taste cues, drag, wave, zoom, pause/reset, food consumption, paint/erase, partial-cue retrieval, ambiguity refusal, empty-cue handling and keyboard drawing.
• All three mocks checked at desktop size and 390 × 844 / 390 × 700 with reduced motion; captures visually inspected. No page errors or unexpected external requests in the suite.
• The scoped Linux/Chromium CI check passed on final commit 9bf25d3 (run). GitGuardian, git diff --check and local Gitleaks scans also passed.
• These checks do not establish biological fidelity, production HDR, real-device touch performance or fun. Production code and deployment were not changed.

[Copilot]

Copilot was unable to review this pull request because the user who requested the review has reached their quota limit.