Random Storytelling in Nodes — v2
Status · SUPPORTED · METHOD UPDATE · July 31, 2026
A working generative storytelling system now creates a fresh monochrome audiovisual composition from procedural node structures, changing motion, selected language, and sound.
Question
Can a browser-based node system generate a new story-like audiovisual composition on each run instead of replaying one decorative loop?
Prior hypothesis
A fresh seed combined with multiple visual families, variable parameters, moving language, and sound would produce materially different compositions while preserving one recognizable editorial direction.
Method
The experiment uses a fresh browser-generated seed, six procedural visual families, randomized composition parameters, layered geometry, and a bounded language library. Words enter as moving typographic actors at changing scale and position. Sound accompanies the composition after the visitor activates it. Full-screen presentation and a standalone downloadable browser screensaver use the same direction.
Direction
Monochrome, procedural, hypnotic, and narrative rather than decorative, with sound.
Observations
- Four verified homepage reloads produced four different seeds, visual modes, compositions, and word sequences.
- Two independent screensaver launches produced distinct rendered frames.
- The full-screen control opened and closed correctly.
- The composition remained responsive without horizontal overflow.
- Music remained visitor-controlled and was raised to a more audible background level.
Evidence
- A fresh cryptographic seed is created for each load.
- Visual family, harmonics, density, layer behavior, word choice, scale, position, and motion vary within bounded rules.
- The public homepage runs the v2 experiment.
- The standalone package passed JavaScript syntax checks, two distinct live renders, and ZIP integrity checks.
- The downloadable package contains no external runtime dependency and does not force audio playback.
Counterevidence and limits
- Random generation cannot guarantee that a seed or visual relationship will never recur.
- Technical variation does not by itself prove that every viewer experiences a coherent story.
- Sound requires visitor activation because browsers restrict automatic playback.
- Performance may vary by browser and device.
- This is a browser screensaver, not a signed native macOS
.saver bundle.
Conclusion
The implementation supports the bounded technical claim: it generates materially different audiovisual compositions from procedural rules instead of replaying one preset loop. The visual language remains recognizable across runs. Whether longer sequences produce stable narrative meaning remains an open research question.
Production consequence
The NOMOTO MEDIA homepage now uses Random Storytelling in Nodes v2 as a live generative feature. The same experiment is available as a standalone downloadable browser screensaver.
Open the live homepage experiment →
Download the Generative Storytelling v2 screensaver →
Next unresolved question
Can the language engine build a beginning, pressure, transformation, and release while remaining genuinely generative?
Return to Transit