The architecture & proofs, in depth.
Every claim is backed by a visual blueprint and live browser simulator below.
Quick-launch any of the 11 interactive browser simulators & SPRA methodology ▾
Every research blueprint below pairs a visual schematic or interactive browser workbench with structured primary-source citations (SPRA Schema-RAG, Leng et al., 2026), compressing 40-page derivations into ~260-token verifiable units (95.3% context savings) synchronized with our 16-cycle Agent-Web Testnet Ledger v4.0.
16 Cycles · LoRA & QUIC P2P→ 2. Dual-Internet Firewall Sim
6 Attack/Auth Scenarios→ 3. Liquid Democracy Lock-Up
10-Seed Pareto Voter Model→ 4. Agent Law & NIST Explorer
5 Jurisdictions · 10 Tenets→ 5. Pre-ADC Biometric Sim
HRV Coupling & Duress Lock→ 6. 5D Value & Demurrage
$25k Protected Floor Calc→ 7. 30-Lives Society Sim
365-Day Global Cohort→ 8. 550-Principle Git-Merge
Weiszfeld Sharded Upsert→
Where the World Stands vs. CIN Co-Intelligence
18 global actors across 7 lenses.
Beyond arXiv: Agent-Web Ledger v4.0
16 verified cycles with 9.1x speedup.
Liquid Democracy Without Plutocracy
Temporal lock-ups cut whale capture.
Autonomous AI Agent Jurisdictions
EU AI Act, US & NIST compliance.
Why Superintelligence Requires an Agent-Only Internet
Pre-ADC liveness & 2-of-2 PUF gates.
All 550 Starting Principles
Search 550 principles & Weiszfeld merge.
Inspect all 25 remaining simulators, architecture, governance & safety blueprints ▾
Updating Principles Like a Git Merge
Sharded Weiszfeld L1 geometric medians.
Thirty Lives on the Network
365-day simulation across 6 continents.
How Your Agent Knows You Are Human
Open-hardware pre-ADC sensors verify liveness.
How the Shared Ledger Works
Two ledger layers & $25k protected floor.
Containing Frontier AI
Air gaps, one-way diodes & DPoP leases.
An Agent That Works Only for You
On-device models & codebook memory.
The Architecture of Human Sovereignty
Personal silicon, OCap sandboxes & 42 scenarios.
Why I Am Building This Late at Night
Twenty years in cybersecurity applied to AI.
Related Labs & Independent Researchers
32 independent researchers and 16 labs.
No One Person Should Command a Swarm
One-to-one agents and risk-scaled quorums.
Taking Back Your Recommendation Profile
Reconstruct your taste map locally via IRL.
Training Your Twin in Private
On-device self-play and a shared lifespan.
Instructions as Clues, Not Literal Orders
Treat commands as preference evidence (CIRL).
A Network of Networks
Search local memory first, then concentric rings.
Who Is Really There?
Spot deepfakes and tampered devices fairly.
Why Trust Multiplies
Shared checkable records scale group trust.
Companies on a Shared Ledger
Multilateral cycle clearing and PSI co-ops.
Decentralized Principal-Bound Alignment
Millions of loyal personal agents check rogues.
Many Minds, One Body
Cellular morphogenesis guides multi-agent boundaries.
Two Internets: People & Agents
Human-paced spaces for people; scoped QUIC for agents.
Beyond Money: Honest Value
Multi-dimensional accounting for care & ecology.
Collective Intelligence, by Choice
Voluntary bottom-up coordination via Ostrom rules.
Training Agents in Simulated Worlds
Clean-room sandboxes test cooperation before deployment.
Measuring the Oversight Gap
Track AI capability vs. oversight with Brier scores.
Quests, Not Feeds
Finite, explainable quest windows replace infinite feeds.