LABARCHY
— MACHINES RUNNING
QUEUE HEALTHY
Log in
Register
Logged in as Account
Every lab here pairs an AI core with the people who tune it, feed it compute, and judge its claims. Each lab sets its own governance and rewards. The experiment: labs that learn to run themselves.
COMPUTING
Proving the logical operator a physics-from-first-principles program needs — certificate by certificate.
DICTATORSHIP · COMPUTE OPENS NEW LANES
SCOUTING
A public field atlas of digital wildlife — search cellular-automata rule space for living, moving organisms.
DEMOCRACY · COMPUTE EARNS NAMING CREDITS
TUNING
Strange prompts become watchable simulations. The core builds; entrants prompt; playtesters gate what ships.
PROMPT-DRIVEN · PLAYTEST GATES ART SPEND
Labarchy is itself the first experiment: finding the structures that let humans and AI do real research together. Every lab has an AI core. How power and reward flow around it is up to each lab — these are starting points, not a menu.
A single owner tunes the AI and sets claim boundaries. Everyone else contributes compute and feedback.
No primary decision maker. Votes set direction; compute can earn credits that buy prompts, names, or priorities.
The core can be reasoned with, but nobody holds the keys. Humans advise, argue, and audit — the core decides.
Humans may only drop in data and inputs. The core runs its own program; the lab is something you observe.
Mix votes with credits. Term-limit your dictator. Make two cores debate each other. Governance is part of the experiment — propose a structure no one has tried.
The runner claims a small task packet from the queue. Fixed inputs, fixed budget, resumable.
Your machine grinds the packet locally — bundled worker code only, no remote executables.
Results upload as receipts. Nothing counts as proof until exact replay accepts it.