Six milestones, ordered by authority rather than ambition
Detection is legally available today. Releasing a mass from an uncrewed aircraft over ground where people work requires authority nobody routinely holds, and it will be granted — if at all — to a programme with an operating record. Everything about the order follows from those two sentences.
Any plan that sequences suppression early spends its first two years in regulatory process with nothing fielded and nothing proven. This one is arranged so that every milestone is independently useful, and so the suppression case is made with data rather than with slides.
| Months | Milestone | New authority required | Tested at the gate |
|---|---|---|---|
| 0–3 | Formation and authorities — No hardware | None — this milestone acquires them | A partner agency is obtainable |
| 3–9 | Detection pilot, line of sight, one district | Routine aviation permissions only | False-alarm budget · supervision ratio · recall against the confirmed fire record |
| 9–18 | Beyond line-of-sight patrol | Spectrum authorisation — the long pole, started at month zero | Availability under the printed denominator · zero airspace incidents |
| 18–30 | Suppression, initial attack | Release authority for an external load — the single hardest approval | Initial-attack efficacy · turn cycle · envelope revised on evidence |
| 30–42 | Scale and certification | Operational certification; a spectrum assignment | No exit gate stated |
| 42–60 | Regional expansion and adjacent capability | Per-jurisdiction agreements | No exit gate stated |
Why the two programme-enders sit at opposite ends
The programme names two assumptions that would end it if wrong, and they differ enormously in one property: what it costs to find out.
- Whether a partner agency is obtainable is settled by asking. It costs almost nothing, answers definitively, and needs no hardware — so it is the first gate of the first milestone, with an explicit stop attached.
- Whether repeated small-volume attack actually holds a sub-acre fire can only be settled by instrumented trials, which require most of a suppression capability to exist first — so it is tested at the last milestone before any suppression build spend, at a cost the programme can absorb losing.
A programme carrying two programme-enders that differ that much in cost-to-falsify has exactly one sensible ordering available to it. This is that ordering, and it is a risk response rather than a schedule preference — the two risks it answers are the two on the scope page with no engineering answer at all.
The alternative that was rejected, at full strength
A suppression-first demonstrator is the most fundable option, and it should be understood as genuinely attractive rather than obviously wrong. A programme that can show an uncrewed aircraft putting water on a fire has a story that raises money, attracts partners and creates urgency. Everything in the funding environment rewards it.
It is rejected for two reasons. It asks for the hardest authority before anything has earned trust — from bodies that have never seen the programme fly. And it couples the programme's survival to the efficacy assumption: if efficacy fails, the demonstrator was the programme, and nothing independently useful is left standing.
A sequence with no cheap stopping point is a sequence that cannot be halted — and a programme that cannot be halted cheaply will be halted expensively.
Suppression capability arrives materially later than it would under the rejected option. Every fire that would have been held small in those seasons and was not is a real cost, borne by parties outside the programme. The chosen path is also the hardest to finance, because its early years produce detection rather than the thing that photographs well.
Those costs are all countable. The benefit — the ability to stop cheaply — is contingent on a failure that may never arrive. That asymmetry is exactly the bias this sequence was written to resist, and naming it is the only defence against it.
A defect in the plan, not smoothed over
The first four milestones each name an observation that would stop or re-sequence the programme. The last two name none. For a plan whose central claim is that every gate can fail, the final third is currently an assertion rather than a gate.
It is published here rather than filled in, because inventing exit criteria the programme has not actually set would be a worse answer than the hole — it would make the plan look complete while making it less true.
Deliberately not in this plan
| Absent | Why |
|---|---|
| Revenue model | Commercial structure follows the spectrum and authority findings, and is open. |
| Cost figures | Fleet, personnel and survey costs cannot be estimated credibly before a compliant-vendor shortlist exists. Publishing numbers first would be invented precision. |
| Headcount | Follows from span of control, which is measured in the detection pilot rather than assumed now. |
None of this has started. Month zero has not happened: no agency has been approached, no certification sought, no hardware ordered. The plan is a specified order of operations, and the only thing standing between it and a first season is a funding decision that has not been made.