SORA prices operational assurance in flight hours. The price is not a policy choice: it is one line of statistics. What is a choice, and what no regulator has written down, is when two operators may add their hours together.
Two operators pool their flight hours to reach a higher assurance level. Who gains more from the arrangement?
With zero failures in n hours the 95% upper bound on the
rate is −ln(0.05)/n = 2.9957/n, which is the Rule of Three's 3/n.
Solve n = 3/λ and the published ladder falls out exactly, one decade of
λ per SAIL. EASA's Annex E footnotes the Rule of Three directly.
| SAIL | λ target | n = 3/λ | JARUS 2.5 | EASA Am.3 | UK CAP 3017 |
|---|
JARUS and CAP 3017 close the route at SAIL V, where the arithmetic asks for 300,000 zero-failure hours — so “not feasible” is a statement of fact. EASA closes it one rung earlier, and nothing in the arithmetic distinguishes 30,000 hours from 3,000 in kind. That ceiling is a regulatory judgement sitting on top of the mathematics.
“The competent authority may accept accumulation of FTB hours between operators if the UAS configuration, operational procedures, training, etc. are demonstrated to be equivalent.”
JARUS SORA Annex E 2.5 §E.3(d)(ii). EASA and the UK carry the same sentence. None of the three defines the demonstration.
The Rule of Three bound 3/n is valid for n draws from
one population with a common rate. Adding two operators and using
3/(nA+nB) assumes they are exchangeable. That
assumption is what “demonstrated to be equivalent” is — not a formality
around the statistics, but the thing the statistics rest on.
If the two populations really are exchangeable this is correct, and is the whole point of pooling. If they are not, the overstatement is the size of the error. No text among the three says how to tell the two cases apart — and a result in the same statistics says the flight hours cannot settle it: two zero-failure samples carry no information about whether their rates differ. The evidence has to be structural.
Zipline's newsroom, 21 January 2026, verbatim: “more than 125 million autonomous commercial miles flown to date”. Flight hours are not reported, so a conversion needs a cruise speed — which is exactly the sort of assumption a conclusion should not rest on. Bound it instead:
| assumed average speed | implied hours | × the 30,000-hour rung |
|---|---|---|
| 100 mph | 1,250,000 | 42× |
| 60 mph | 2,083,333 | 69× |
| 30 mph | 4,166,667 | 139× |
Across a 3.3× span of plausible speeds the answer stays between 40× and 140×. The conclusion does not depend on the assumption. The largest operators are not held back by the ladder; they cleared it long ago. So pooling is not how anyone reaches the top rung — it is how an operator without hours comes to stand on the record of one that has them, at the exchange rate panel 2 computes.
And the quantity the rule is written in is not the one the operators report. Checking the three largest publicly: Zipline reports miles and deliveries, Wing reports “1M+” deliveries, Amazon reports packages. None of the three reports flight hours. So for those three, which side of panel 2 they are on is not a fact an outside party can check from what they publish.
Sources, all primary and quoted verbatim: JARUS SORA Annex E, JAR-DOC-28, Ed. 2.5, 13.05.2024 · EASA AMC & GM to Reg. (EU) 2019/947, Issue 1 Am. 3, ED Decision 2025/018/R · UK CAA CAP 3017, July 2024 · UK Regulation (EU) 2019/947 library, GM1 Article 11 Annex E (GM.FTB), ORS9 Decision No. 46, p.485, July 2026 · Zipline and Wing company pages. This page performs arithmetic on published figures and quotes regulatory text; it is not legal or airworthiness advice.