The first claim failed.
Not supported210 calls; relevant evidence moved three complete units, but nuisance movement and uneven observability defeated the frozen criterion.
Review Experiment 1 →Complete public lineage
Every experiment remains reviewable.
The newest instrument is front-facing, but its predecessors and their negative results remain part of the evidence.
Scale did not restore identification.
Insufficient evidence888 resolved coordinates; both provider arms returned insufficient evidence, with 66 of 288 matched units fully valid.
Review Experiment 2 →Conservative, stress-tested, and locally executed.
No frontier-provider scienceThe governing SR-8R1 campaign scheduled 1,600 runs across Qwen 2.5 14B and Gemma 3 4B. 823 rows reached a non-transport terminal state—822 substantive assessments and one measurement-unavailable row—while 777 were transport-unavailable. A separate 4,800-run Qwen campaign remains earlier stress evidence and is never pooled with this result.
Review Experiment 3 →