The moat of a reasoning platform is its calibration track record against real outcomes. Open-sourcing the engines hands competitors the code but not the record, which is backwards. ARIAA publishes methodology (gated whitepapers, hosted dashboards, rate-limited public API); it does not publish implementation.
Related
- Calibration (Brier Score) — The discipline of grading probabilistic forecasts against realised outcomes. Low Brier score = well-calibrated.