OASI/AERA: Operational Artificial System Intelligence Research Preview

Abstract

This archive is the aggregate OASI/AERA v0.2.1 research preview. Operational Artificial System Intelligence (OASI) is the canonical name of the research paradigm; organismic computing describes its architectural hypothesis, and AERA is the bounded assurance mechanism implemented here. Operational denotes system operation, not production readiness, and the name does not claim achieved general or superintelligence, consciousness, deployment, external validation, or superiority. The archive contains the unchanged bounded Rust AERA reference runtime and tests, the v0.4 preprint and sources, specifications and claim boundaries, deterministic S5/S6 fixture data, Linux/WSL-relocatable runners qualified on WSL1 x86_64, detached aggregate checkers, tests, manifests, an SBOM, a license inventory, and reproducibility documentation. The unchanged historical Rust crate remains versioned 0.1.0-research-preview; v0.2.1 identifies the aggregate research release. The package preserves 2,700 deterministic records across 90 cells. S5 is a negative result: it did not establish an OASI advantage over the cooperative idempotent B3 receiver. S6 diagnoses a retry-versus-no-retry duplicate/omission tradeoff under a non-cooperative fixture; it does not isolate an OASI-specific mechanism advantage. Repetitions are implementation-stability traces rather than independent population samples. Fault labels denote simulated control-flow traces, not physical power loss, process termination, severed transport, or storage tearing. This release is local, fixture-only research evidence. It does not establish a complete operating system, universal effect guarantees, production or security certification, performance superiority, real-world deployment, external replication, or general superiority. Development and internal adversarial review were extensively AI-assisted and project-controlled; they are not human peer review, independent replication, certification, or institutional evaluation. Licensing is path-specific and is recorded in LICENSE_INVENTORY.json.