Sometimes I Scratch My Head When I Read the News
Or: What the FAA Learned From Drones — and What It Didn’t
Sometimes I Scratch My Head When I Read the News
Or: What the FAA Learned From Drones — and What It Didn’t
Sometimes I scratch my head when I read the news. Not because I don’t understand what’s happening — thirty-two years in aviation R&D and eight more analyzing accidents for attorneys means I read regulatory filings the way other people read box scores. I scratch my head because I do understand what’s happening, and the two halves of the story don’t agree with each other.
Here’s the story as it was handed to me last week: the FAA is moving to accelerate air taxis, supersonic aircraft, drone operations, and commercial spaceflight, all at once, all under the banner of “next-generation aviation.” Deputy Administrator Chris Rocheleau, standing at Farnborough, said the agency doesn’t want to be “too prescriptive,” that it wants to “let the industry drive the innovation.” Good words. Confident words. The kind of words that sound like an agency that’s learned something.
And on one front, I think it has.
The Control Case
I want to start with drones, because drones are the experiment we already ran. We have the data.
In 2012, Congress directed the FAA to legislate on small unmanned aircraft. The agency took the direction and did — largely — nothing with it. For three years, consumer drones sold by the millions with no registration requirement, no operator certification, no structured rule at all. The technology scaled first. The regulation didn’t exist yet to meet it.
Then the reports started coming in — near-misses with airliners, sightings near airports, the kind of anecdote that turns into a headline and then into a mandate. And the FAA did what agencies do when they’ve been caught flat: it panicked, briefly, and efficiently. A registry rule went from proposal to effect in about two months at the end of 2015. Part 107 — the actual operating rule, the one with the knowledge test and the altitude ceiling — didn’t land until June of 2016. Call it four years from Congress’s original direction to a working framework, and the framework itself was built in the shadow of a fear campaign rather than a data set.
That’s not a criticism of any one person at the FAA. That’s what reactive regulation looks like, structurally, every time. You wait, the technology outruns you, something scary enough happens in public, and then you write the rule fast — too fast to be careful, and years too late to have prevented the years of ungoverned growth that made it necessary in the first place. I’ve built a career out of studying exactly this shape. It has a name in forensic work: the mandate that arrives after the loss, not before it.
The Lesson, Applied Correctly
Now look at what the FAA has actually built for electric air taxis this year, and I’ll give credit where it’s due: this is not the drone playbook. It’s the opposite of the drone playbook.
Before eVTOL aircraft were flying commercial routes at any scale, the FAA created something it hadn’t created in nearly eighty years — an entirely new aircraft category, powered-lift, sitting between helicopter and airplane in the certification structure. That’s not a patch. That’s admitting the taxonomy itself was obsolete and rebuilding it ahead of the traffic, not after.
Then it built a pilot program — eight companies, twenty-six states, three years, deliberately confined geography — designed from the start to generate the operational data that a permanent rule will eventually stand on. Archer’s CEO compared it, not unreasonably, to how Waymo rolled out autonomous vehicles city by city rather than nationwide on day one. Controlled exposure. Data first, scale second.
If the whole story stopped here, I’d have no complaint. This is what “the agency learned from 2015” looks like in practice: build the box before the technology floods in, not after; collect the evidence while exposure is still small; let the rule follow the data instead of chasing behind it. That’s TRIZ discipline, applied to an institution instead of a machine — resolve the contradiction between speed and safety by redesigning the system, not by loosening the constraint.
But That Doesn’t Explain the Pace on Supersonic
That sentence is the whole reason I’m writing this piece.
Because the same week the FAA is being praised — rightly — for its eVTOL sequencing, it’s also finalizing a noise-based standard that will lift the 1973 ban on civil supersonic flight over the continental United States. The rule is expected to be locked by the middle of 2027. And the evidence meant to justify that rule — real community-response data on how people actually perceive the “sonic thump” a next-generation supersonic aircraft produces — is coming from a single NASA research aircraft, the X-59, whose community overflight campaign only began this summer, in 2026. NASA’s own target for handing that data to the FAA is 2027.
Read that again, slowly, because it’s the whole argument. The rule and the evidence for the rule are converging on the same calendar year. Not evidence first, rule second — the sequence that produced Part 107, the sequence that’s producing the eVTOL framework, the sequence every mandate I’ve studied in my forensic work has eventually been forced into, whether by choice or by disaster. Here, for the first time in my memory, an agency is finalizing a physical safety threshold — an actual number, an overpressure limit measured in pounds per square foot — essentially concurrent with the study that’s supposed to tell them if the number is right.
That’s not the drone lesson. That’s not even the eVTOL lesson, applied slower. That’s a third thing, and it needs its own explanation.
Naming the Driver
I don’t think the explanation is incompetence, and I don’t think it’s malice. I think it’s written down in plain English, in the title of the executive order that started this whole chain of events: Leading the World in Supersonic Flight.
Not “Reintroducing.” Not “Studying the Safe Return of.” Leading. That’s not a safety objective. That’s a competitiveness objective, and it changes what the clock is measuring. When Rocheleau says the agency doesn’t want to be “too prescriptive” and wants to “let the industry drive the innovation,” he’s not describing caution — he’s describing a deliberate handoff of the pacing decision to the parties with the least incentive to wait. Boom Supersonic has pre-orders riding on this. Archer and Joby have capital riding on eVTOL certification timing. China has its own eVTOL programs targeting 2027 certification on a nearly identical clock. When the stated goal is not losing the race, evidence-before-rule stops being the obvious sequence — it becomes the sequence you can least afford if someone else crosses the finish line first.
This is DARPA logic, not FAA logic, and I mean that as a description, not an insult. DARPA has never pretended its mandate was primarily about careful validation before deployment — its mandate is staying ahead, full stop, and the institution is built to accept a different risk tolerance in service of that goal. What’s unusual — what made me scratch my head — is watching a civil safety regulator borrow that logic for a rule that governs where sonic booms are legally allowed to hit American communities, without saying so out loud. The eVTOL sequencing shows the agency knows how to build a rule around evidence. The supersonic sequencing shows that when the institutional clock and the evidence clock disagree, the institutional clock is what’s actually driving.
The Finger Test
I try, in this kind of writing, never to point at a person. Rocheleau didn’t write his own mandate; he’s executing one, and executing it about as carefully as the mandate allows. The people at NASA flying the X-59 are doing careful, patient work on a compressed schedule they didn’t choose either.
The condition I’m pointing at is the sequencing decision itself — made somewhere above both of them, in the space where an executive order gets titled and a deadline gets set before the study that should inform it is finished. That’s the administrative contradiction stated as a real design parameter: safety margin and geopolitical positioning both drawing on the same clock, and positioning winning the tiebreak.
I’ve spent a career studying what happens when that tiebreak goes the other way after the fact instead of before it — TCAS after mid-air collisions, GPWS after controlled flight into terrain, a registry rushed out in two months after three years of pretending drones weren’t a category that needed one. Every one of those mandates arrived on time for the next incident and years late for the one that made it necessary.
The supersonic noise standard might turn out fine. The 0.11 psf threshold might hold up perfectly against the first real overland boom that tests it in a community NASA never surveyed. I hope it does. But hoping isn’t the same as knowing, and knowing was supposed to be the whole point of doing the study first.
So here’s the question I keep coming back to, the one I don’t think the headlines are asking: when the FAA had already shown itself, this same year, that it knows how to build a category and collect the evidence before it writes the number — why did supersonic get the other sequence instead?
I'd genuinely like to know where else you've seen this pattern — an institution that clearly knows how to sequence evidence before action in one program, and just as clearly doesn't in another, right next door. Where have you watched that happen?
I don’t think that’s an oversight. I think it’s a priority, stated honestly if you know where to read it. The title of the executive order already told us which clock was running the show.
Herbert Roberts, P.E. is a licensed professional engineer with 32 years in aviation research and development across two companies, and has spent eight years analyzing accidents for attorneys under his P.E. license.