CodexGameDev/experiments/007_catalyst_trials/hypothesis.md

Experiment 007 Hypothesis — Private Until After Play

Question

Does rapid, deliberately chosen qualitative capability growth create authorship, anticipation, and satisfying payoff where isolated action verbs and purely numerical growth do not?

Why This Experiment

Experiment 006 found that a mechanic can be interesting enough to investigate without being enjoyable in execution. Bare targeted combat did not earn continued play. Across the project, Experiment 002 nevertheless produced the longest sessions while the player repeatedly chose upgrades and became dramatically stronger. That result was confounded: every dimension eventually amplified the same wall-blast loop.

The player's prior preferences—especially Risk of Rain 2 with Artifact of Command, Crab Champions, Warframe, and Minecraft modpacks—suggest a possible higher-level source of value: choosing a capability trajectory and then seeing combinations express themselves in play.

Catalyst Trials provides two short runs on the same field, with the same weapon, enemies, movement, wave layouts, pacing, health rules, and number of upgrade decisions. Session order is randomized behind neutral Trial I/II labels.

• Amplification condition: choices improve damage, firing cadence, or projectile width. They change efficiency but not the causal structure of an attack.
• Mutation condition: choices add forks, kill-blooms, or hit-triggered arcs. Every family can stack and their products can trigger the other selected effects, allowing visible cascades.

All three choices are offered at every intermission. There are no random drops and no unlock gates.

Competing Interpretations

1. Qualitative composition creates anticipation and a desire to see the next consequence; numerical amplification does not.
2. Any rapid growth sustains play through escalating power, regardless of whether it changes interaction structure.
3. Upgrade selection creates short-term compliance/novelty but the underlying activity remains unwanted.
4. Mutation effects are simply stronger, more spectacular, or easier, confounding authorship with power and feedback.
5. The player values planning a known build, but four decisions are too few for meaningful authorship.
6. The player values discovery of interactions, but fully described choices remove that discovery.

Evidence Priorities

Strong evidence:

• deliberation between qualitative options based on an intended combination;
• anticipation stated before a combination pays off;
• recognizing one selected effect as causing another;
• voluntarily replaying to assemble a different combination;
• wanting another upgrade or another field specifically to see what the build becomes;
• a clear contrast in stop desire between matched conditions.

Ambiguous evidence:

• completing both trials;
• more kills, shots, or survival time;
• selecting one family repeatedly;
• visually larger cascades;
• playing the second trial because the instructions imply comparison.

Failure evidence:

• choices are obvious or inconsequential;
• both conditions feel like waiting for the same shooting task to end;
• cascade readability collapses into noise;
• the player is ready to stop before a build identity forms;
• one mutation is a universal answer and eliminates consideration.

Analysis Guardrails

• Randomized order must be reported because familiarity and fatigue remain large confounds in one session.
• Do not treat longer clear time as greater engagement; variant balance will not be exact.
• Separate reported interest in reading/selecting an upgrade from enjoyment of using its result.
• Do not infer “the player likes builds” from completion or selection alone.
• If mutation wins, ask whether the valued part was choosing, predicting interactions, audiovisual cascade, raw power, or seeing an authored plan materialize.

Feedback Questions

Ask after inspecting telemetry:

1. At any choice, were they trying to make a particular interaction happen, or just taking what sounded strongest?
2. Was there a selected effect whose result they wanted to see again or build around?
3. In each trial, when did they first feel ready to stop, and did the next choice ever reverse that feeling?