Conceptual map only — 36 nodes · 41 edges · Alpha Factory → Production Trader influence BLOCKED.
Alpha Factory V3
Public Research Lab
Systematic research with institutional memory. A kill means the idea was falsified — a legitimate research outcome, not a system error. Research does not auto-control live capital · 0 survivors promoted · Campaign 001 not run.
30 experiments evaluated · 7 falsified · 3 families preregistered (draft locked, not eligible for promotion). Interactive sections load from the public research lab export. Source as-of is shown separately from export regeneration time.
Level 1 · 5-second overview
Level 2 · Explore (30–90s)
Experiment families
Falsification wall
Public-safe killed families — mechanisms summarized, no thresholds or formulas.
Novelty taxonomy
G/H/I preregistration (draft locked)
Opportunity graph
Shadow intelligence
Missed-winner gate stories and virtual rankings are SHADOW counterfactuals — not verified edge. Full surfaces below (moved off homepage in P8).
Missed Winners
Candidates that moved after a gate held — SNAPSHOT/ARCHIVE labeled from source as-of, never export clock. Not trade signals.
Daily Discoveries
Intelligence cards from research exports — freshness from source events.
Intelligence Timeline
Chronological research pipeline events.
Shadow Pattern Rankings
Virtual win-rate rankings — SHADOW class, not trade signals. See epistemic taxonomy.
Level 3 · Technical deep dive (public-safe)
Pipeline stages
- Institutional memory — prior FALSIFIED registry imported.
- Novelty engine — NEW / VARIANT / DUPLICATE classification.
- Cost & null gates — after-fee falsification.
- Walk-forward — sealed holdout never reused.
- Influence gate — live_influence=false.
- FALSIFIED: Preregistered gate failure — legitimate research outcome, not system error.
- survivors=0: Honest result after cost/null/WF gates — factory working as designed.
Export as-of pending JavaScript · static tier-0 valid without JS