For long-lived agents, keeping state is not enough: the system must know which branch head is authoritative and how it was authorized.
Why it matters: The design makes authority and freshness part of state activation, rather than assuming that retained state is trustworthy. The reported verification supports the protocol within a bounded executable model, not an independently audited implementation.
The captured HN page gives a story title, not the linked article. Its partial comments discuss a claimed cross-model trace-replay technique and disagree about how to frame it.
Why it matters: For anyone assessing trace confidentiality, separate the proposed replay mechanism from commenters’ workarounds, legal opinions, and anecdotes. This capture cannot verify the linked story or establish reproducibility.