Soft Kill
Not every threat needs a missile. The system learned to jam — to deny a drone's control link and let its own failsafe bring it down — and to recognize when jamming won't work at all.
Jammers fail. Hardware fails, physics fails, the adversary adapts. So jam failure is not an edge case; it is the case. The system models it as an explicit condition, not a probability: when a committed jam does not suppress, the threat re-enters the queue with reason, and the next effector is drawn from the same scarce magazine.
The degradation is monotone and measured. At zero injected failure, protection holds at 96.7%. As failures rise through 10–30%, it steps to 91.7% and holds — the plateau is quantization and jams that were declared but never committed. At 50%, it steps again to 88.3%. The system does not collapse. It degrades along a curve it has already walked.
Sim-validated on a 60-threat scenario set. Not field-proven.
60 THREATS5 MEASURED POINTSSIM-VALIDATED