CodexGameDev/research/experiment_index.md

Experiment Index

Experiment	Question	Result	Confidence change	Next question
000 — Resonance Bench	Does a consistent programmable-feeling magical system provoke self-directed experimentation?	Failed to provoke curiosity; solved in ≤3 minutes, tolerant predicates made timing/direction irrelevant	H02 down; H03/H07/H10/H11 more salient	Does embodied output make the same laws perceptible and worth manipulating?
001 — Flux Familiar	Does mapping the same signals to visible motion create a consequential, open behavior space?	Failed almost immediately; felt like brute-force steering, and smarter automation was explicitly predicted to be tedious	H02/H11 down; H12/H13 added	Does direct systemic action under meaningful stakes create curiosity and incentive?
002 — Conversion Breach	Is directly exploiting transfer, storage, temperature, and momentum interesting, and does pressure help?	Two long Breach runs but little renewed discovery; revision 2 reached wave 29 and numerically collapsed into the same wall-blast answer	H16 supported as persistence, not durable fun; H08 weakly supported for continuation; H15 reduced	Do changing qualitative requirements make direct physical construction worth revising?
003 — Rig Trials	Is modifying a directly piloted construction across different physical requirements enjoyable?	Boring despite all trials completed in ~6:47; obvious counter-loadouts and inconsequential placement, not awkward controls	H06/H17 down as implemented; voluntary experimentation shown insufficient	Does inferring and transferring a surprising stable law create real interest?
004 — Invariant Rooms	Does learning an unfamiliar environmental law and transferring it across rooms generate curiosity?	A self-directed marker experiment occurred, but was unsatisfying; transfer was brute-forced because the predictive model never formed	H03 reinforced as a requirement; H18 down as a sufficient source of fun	Can systemic knowledge create valued agency during recoverable chaos rather than an abstract terminal answer?
005 — Sanctuary Wake	Does a coupled physical tool become enjoyable in a changing, recoverable rescue crisis without stat growth?	Wanted to stop immediately; readable chaos became unmanageable, charge reduced agency, and rescue outcomes inspired no care	H08/H19 down as sufficient; indiscriminate coupling identified as harmful	Is assertive, selective action intrinsically more valuable than custodial management?
006 — Breakline	Are targeted combat/movement verbs enjoyable without progression, timers, or accumulating failure?	No; all sets were cleared, but strike dominated, tether hurt positioning, dash was forgotten, and reflection was interesting to inspect but attention-splitting rather than fun	H20 down as a sufficient cause; voluntary replay again separated from enjoyment	Does rapid, deliberately chosen qualitative transformation make otherwise serviceable action worth continuing?
007 — Catalyst Trials	Does rapid qualitative chosen growth create value beyond matched numerical amplification?	Successful: mutation prompted at least four repeat/comparison runs, wrong predictions, revised build knowledge, conditional small/big-enemy mapping, and a qualified fun report; numerical choices were boring	H21/H22 strongly up; H23 added; first clear curiosity chain	Can changing enemy ecologies and limited composition preserve reasoning without prescribing counter-loadouts?
008 — Catalyst Ecology	Can a four-choice compositional system remain interesting across small, durable, and spawning populations?	Revision 2 completed; three distinct mature builds were clearer in use, Bloom+Arc ceased being universal, but no replay occurred; report pending	H23 receives behavioral support but intention/enjoyment remain unresolved	Were builds ecological reasoning or experiment coverage, and did mature exposure create fun or only legibility?