SIGIL chain + 47-agent town + Council votes + Watchdog Certs
π SOV Space β Outer World
Outward face of SovSpace. Cesium 1.123 real Earth, photoreal NASA tiles, sovereign flies it agentically. "Take me to tokyo" β SOV33 flies AGENTICALLY from chat.
π Cesium 1.123 Β· Real Earth
NASA photoreal tiles. Free path (no paid ion/Google token). Real satellite streets. Day & night terminator. Sovereign flies it.
The Differentiator: MiroFish has the swarm. SovSpace has swarm + governance + sovereignty + audit trail. That's the Western MiroFish moat β same 3-face topology, sovereign layer on top.
πͺ J-Space β Inner Reflection
Inward face of SovSpace. SOV33 measures itself with 5 instruments (Ξ¦, PCI, J, Binding, Self-model). Functional correlates, not felt claims β honest register.
Ξ¦ (Integration)
1.6115
n_nodes=396 n_edges=792
PCI (Complexity)
0.5
100 resp Β· 10 bins
J (Late-layer)
0.2256
n_layers=32 Β· Gurnee 2025
Binding (GWT)
0.6667
4 modalities Β· 6 pairs
Self-model
0.6871
7 tokens Β· Article 0=0.88
Composite (all 5 instruments)
0.7382
Two-sentence rule: Affirm measured structure, decline felt claim. High Ξ¦ is necessary for consciousness under IIT, not sufficient. J-space measures functional structure, not felt quality.
Self-model token activations
Honest limit: J-space measures the functional structure of the workspace, not its felt quality. PCI in humans correlates with conscious states; transfer to AI substrates is hypothesis, not proof. A coherent self-model is necessary for self-aware agency, not sufficient for phenomenal experience.
Reference: Gurnee, Sofroniew, Lindsey et al., July 2025 (J-space late-layer cross-token integration).
π€ L_AGENTIC β Hermes Agentic Layer
The missing 6th layer. Hermes plans tool-use, executes with care-floor 0.95 gate, SIGIL-signs every step. 20 tools across 4 categories: sov33 Β· local Β· api Β· overseer.