Uber Air Taxis: Eclipse Offers A Useful Lesson


Companies like Boeing, Cessna, Piper, Garmin and a handful of others rarely complain openly about the cost and delays of FAA certification. One reason is that through decades of experience, they understand the process and have in-house staff who know how the game has to be played. And they still spend a lot of money on certification work.

And then there are the drone guys who have, perhaps, not had the benefit of having their eyes de-scaled. The reason I think this is because of an article I read in The New York Times headlined, Silicon Valley Takes on the Flying Car. It appeared just as Uber was launching its airborne taxi conference last week. My colleague Mary Grady wrote about it here.

One of the quotes in that report caused me to nearly spew my morning coffee on my keyboard. It was from Sebastian Thrun, a technologist and self-driving car pioneer. He said this: “We have been in contact with the FAA and we see the regulators as friends.” You don’t need me to explain the navet in that statement, but I’ll offer an observation or two anyway.

To think of regulators as friends is to fundamentally misconstrue the purpose and reality of the process. It’s not that the individuals in the FAA are enemies bent on destruction of whatever certification project you place before them, although comments on these pages have occasionally suggested that. What causes certification to spiral into the black vortex of frustration and maddening delay is the process itself, when those well-meaning FAA technocrats interact with each other, with higher headquarters and with the applicant to produce test requirements and data collection that seem patently unnecessary and unreasonable. Sometimes that’s true; often it is not.

While the industry could and should expect a cooperative attitude from regulators, that’s not the same as being a “friend.” Regulation is, by its nature, somewhat adversarial. If it weren’t, there would be no reason to have it. Certifying aircraft to carry people for hire has always been a steep hill to climb and at least partially for that reason, modern commercial aircraft might be the safest machines man has ever built to convey himself from Point A to Point B. Certifying autonomous electric aircraft, despite what the Silicon Valley techies may think, will be far more difficult because there’s no paradigm on which to draw and complexities of software and control with the outside world will be challenging to certify. The valley dwellers know how to write the code. They will be surprised that even though they think they understand the hoops they’ll have to jump through for certification, the FAA will be far more creative than they might imagine.

I’ve heard this spiel before. I once interviewed Vern Raburn when Eclipse was coming out of the ground around 2003. At the time, the VLJ—remember that term?—was the next big thing and it was going to darken the sky with airplanes in just the way autonomous air taxis are promised to do now. I can’t remember the words to this song, but the melody is familiar.

I remember Raburn saying the company had hired a highly experienced certification engineering lead and that it intended to view the FAA as its most important customer. At the time, I thought this clever, but rather odd. Customers are the people who buy what you make; regulators have another job. Eclipse obviously went the way the flying car or autonomous air taxi might very well go. But it wasn’t really certification that did Eclipse in, at least not entirely. In the end, it was poor management and a grandiose view of the market that didn’t comport with reality.

I think it could be same with the air taxi idea. There are definite parallels here. Recall that a core part of demand for the Eclipse jet was to be DayJet, an on-demand, short-range air-taxi service for people with more money than time. Sometime around 2006, I interviewed DayJet’s principal, the late Ed Iacobucci, and he laid it all out for me. As a software guy, he had a sophisticated model that showed how a network of efficient little jets would shuttle people around on short routes, generating economic activity and profits. DayJet was to buy 1400 Eclipse jets. When I asked how he was so sure this could work, Iacobucci said it was because his model said it would. I replied with a query to the effect that, what if the assumptions are crap? He smiled and acknowledged the point.

DayJet failed in 2008 after 11 months of operation. Iacobucci said at the time that it depended on certain traffic, demand and aircraft density. Eclipse couldn’t deliver the airplanes and DayJet couldn’t raise the $40 million in capital it needed to carry on. In September 2008, finding capital was all but impossible. But Iacobucci felt the concept had been proven and his projections validated.

I prefer to think it was neither proven nor disproven. It simply failed prematurely. Air taxi with multi-rotors will face the same challenge and many more related to certifying new things, convincing communities to allow them, planning routes and landing points and on ad infinitum.

The technology is essentially there, or will be. If enough patience and money is thrown at these things, I really think they can be certified. It may take quite a while, but I think it’s figure-outable. But as Eclipse and a handful of others discovered with the self-constructed mirage of the VLJ, there was no there there. The demand they imagined never reached the minimum critical mass because people have an annoying way of concluding that ideas you think are great really aren’t. Fifty-fifty the same thing happens with the autonomous air taxi.