Sovereign Takeoff Engine (STE): Observable-Only Supergrowth Laws for No-Meta Autonomous Intelligence

Abstract

This preprint specifies observable-only supergrowth laws for no-meta autonomous intelligence under deterministic replay and fail-closed certification constraints, yielding auditable capability-acceleration criteria without privileged external judges. It separates artifact-verifiable governance from optional stochastic validity layers and formalizes ledger-anchored progress credits, lineage accumulation bounds, anytime-valid e-value testing, and conservative physical feasibility envelopes.

Keywords

• no-meta autonomous intelligence
• observable-only governance
• supergrowth laws
• deterministic replay
• fail-closed certification
• audit ledger
• e-values
• anytime-valid sequential testing
• filtration constraints
• capability acceleration
• physical feasibility envelopes
• AI safety verification

Identifiers and source records

• DOI
• Works entry
• ORCID