Independent rescue
The study separates the rescue signal from the main uncertainty path so its contribution can be tested rather than assumed.
Phase 11 · P14RSimulation-only research · current frontier · Phase 11 P14R
Evidence before confidence.
AegisLand studies safety-supervisor behavior under frozen simulation protocols. The current result keeps the successful evidence and the failed protected gate visible in the same place.
Protected availability98.53%
Lateral 95% coverage96.17%
Altitude 95% coverage95.82%
Rescue recovery94.63%
Research progression
Every phase stays reachable.
Current frontier
A mixed result is still a result.
Phase 11 P14R passed the required seen-transfer gates and preserved strong protected coverage and availability. One locked protected component failed, so the study closed without turning a near-pass into a safety claim.
Robust uncertainty transfer
Coverage is evaluated under a frozen protocol instead of retuning after protected results become visible.
Frozen evaluationVisible failure boundary
The lateral tail-efficiency miss remains part of the result and prevents final confirmation.
Safety acceptance: falseResearch method
Inspectability is part of the product.
AegisLand keeps protocols, evidence, frozen predecessors, failed gates, and source links visible so the website communicates what the experiments actually support.
Freeze before protected evaluation.
Research choices should not quietly move after the protected result is known.
Keep limitations beside metrics.
Strong percentages do not erase a failed preregistered gate.
Keep the lineage navigable.
Earlier phases remain available instead of being rewritten to match the newest narrative.
Research archive
Start with the frontier. Trace backward when you need the evidence.
Phase 11 contains the current protected-validation story. The archive keeps the earlier system changes and evidence chain accessible.