Experiment 000 — Resonance Bench

Question

Does interacting with a consistent, programmable-feeling magical system create enough curiosity that the player voluntarily asks and tests their own questions?

Hypotheses

The primary hypothesis is that phase, timing, conservation, storage, and feedback can form a legible curiosity chain. Competing explanations are that abstract engineering needs world consequences, that circuit-like manipulation is not intrinsically compelling, or that interface friction masks the idea. See hypothesis.md.

Difference From Previous Experiment

This is the baseline experiment.

Expected Result

Strong support is not merely finishing the three observations. It is changing a working network to answer a new question, especially after discovering resonance, cancellation, or timing behavior. A clean stop after prescribed objectives, boredom after understanding the laws, and interface frustration imply different next experiments.

Controls

No time pressure, enemies, unlocks, randomized parts, progression, or world-fiction consequences.
Every operation is available from the beginning.
The three fixed wells provide the same signals on every run.
Objectives observe behavior and do not award capabilities.

How To Run

From this directory:

./run.sh

Then open http://localhost:8000. The prototype also works by opening prototype/index.html directly, although serving it avoids browser-specific local-file restrictions.

What To Pay Attention To

After you satisfy an observation, do you want to alter the network anyway?
Did any result make you ask a new question about the laws?
Which moments felt like reasoning, and which felt like wiring work?

When you are done—even if boredom arrives quickly—use Export JSONL, then report what happened. The most useful first feedback is: when you wanted to stop, one thing you tried that was not required (if any), and what you wanted to try next (if anything).

Result

Playtested on 2026-08-16. The player completed all three observations in roughly three minutes or less and felt no reason to continue.

No voluntary experiment followed completion.
The player did not notice that sources were cyclic or that flux had defined direction.
The observation predicates were permissive enough that arbitrary timing and direction would still produce competent solutions.
A merely working solution felt fully sufficient; there was no meaningful reason to improve, vary, or interrogate it.

No JSONL was available for this interpretation; the player's direct report is decisive for the primary behavioral question.

Interpretation

Experiment 000 did not produce the desired curiosity chain. This is a failure of the experiment as built, not yet a clean rejection of programmable magic.

The observations failed to make the central laws causally relevant. They measured loose occupancy and threshold events rather than behavior whose outcome depended on phase and timing. Consequently:

completion is not evidence that the player formed the intended mental model;
lack of post-completion optimization is unsurprising because improvement had no consequence;
cycles and vector direction were implementation details rather than perceived phenomena;
the prototype's expressive space appeared exhausted as soon as its three binary predicates were satisfied.

This lowers confidence that an abstract signal bench is intrinsically compelling in its current form. It increases confidence that legibility must come from visible consequences, not explanatory text, and that objectives must discriminate between structurally different outputs if they are meant to expose a law.

Next Best Experiment

Build a controlled successor that retains deterministic cyclic sources and transformations but maps output vectors onto visible motion in a small world. Direction becomes literal direction; timing changes trajectories; cycles create observable repeated behavior. This tests whether embodied consequence makes the same knowledge interesting, versus the competing explanation that the signal substrate itself is too limited.

4.1 KiB Raw Permalink Blame History