The Friction-Aggregation Policy Channel
Question. Does the network's field-friction signal surface structural problems the institution cannot see from the inside, or is it just articulate complaint?
Analogy. It works like a flaw you can see only as a colour, in a room lit by a single-wavelength sodium lamp: under that one light every colour collapses to the same shade, so the flaw is invisible to everyone in the room no matter how sharp their eyes, because the light they all work under carries no way to register it, and only someone who carries the part outside into daylight ever sees it. Internal review is everyone in that sodium-lit room, working under the one light it cannot question. The study asks whether the network's outside-the-room signal really surfaces structural problems the institution cannot see, or whether it is just articulate complaint it could have reached on its own.
What's at stake. The proposal pitches the network's policy-signal channel as a distributed early-warning sensor for DND: forward-deployed operators send anonymised reports of where a process contradicts its own stated objective, and the aggregate is supposed to reveal structural problems no single report shows. The load-bearing and untested claim is epistemic: that this outside-the-apparatus signal sees things internal audits structurally cannot, because every internal audit runs under the rules of the apparatus it is auditing, so the questions that implicate the rules themselves never get asked. If that is true, the channel is a genuinely new sensor. If it is false, it is a high-volume complaint stream with good public relations, and DND should not build it.
The two answers it decides between. Either operators working inside real friction but outside the institution's incentives surface structural problems internal review cannot, because the insider audits under the rules it would have to question, so the aggregate carries signal no internal instrument produces; or the aggregate contains nothing an institution's own review could not reach, the outside-view advantage is illusory, and the channel is noise with a credible label. A blind comparison of what the network surfaces against what internal review surfaced over the same period tells them apart.
What a null result would mean. If the network aggregate surfaces no structural problems beyond what internal review reached, the distributed-sensor claim needs revision: the channel is articulate complaint, not signal, and the policy-channel half of the dual-use case should be dropped. That is a finding about whether outside-the-apparatus observation has unique epistemic value, not a sign the platform was built wrong.
Why this matters to defence. A distributed sensor that surfaces structural problems the institution cannot self-see is directly valuable to a department that runs on internal review (DRDC Objective 4; the institution and force development). It changes a concrete decision: whether to build and trust the policy-signal channel at all. Two conditions ride on the answer: whether each surfaced problem can be routed to the lowest official with authority to act on it, and whether the channel can be kept from being spoofed or poisoned, the serious risk that a trusted sensor becomes an adversary's laundering route into government. The reception question, whether institutions act on a disinterested signal that implicates them, is the separate downstream study; this asks the prior question, whether there is real signal to act on.
How we would run it. Pick a domain where DND runs its own internal review and the network also generates friction reports, Northern procurement being the natural fit given the network's Ranger and remote-operations operators. Collect the network's aggregated friction signal and the institution's internal-review output over the same period, strip both of any mark of source, and have an expert panel rate each for whether it identifies a real, structural, actionable problem. The discriminating measure is whether the network aggregate contains structural problems the internal review missed and that the institution could act on, beyond articulate complaint. The routing condition is measured alongside: can each surfaced problem be mapped to a lowest authority who could act. The only-us part is operators working inside real friction but outside the institution's incentives, which a managed cohort, being inside the apparatus, cannot reproduce.
Earliest start. Stage 15: the study needs the policy-signal channel and a live internal-review domain to compare against.