December 2024
Well, I appreciate you are trying to have a bit of fun around a pretty serious subject and that’s fine by me. But, if you seriously think single piloting isn’t going to happen then I’d say it’s probably time to retire (oh… you have already!).
You conveniently fail to mention that humans are often quite fallible even when there are 2 present trying to resolve a bad situation - and even crash aircraft deliberately. In fact (as I am sure you know) the majority of aviation calamities are caused by human error.
So, one pilot is one thing: what’ll the Luddites be saying when the inevitable happens and a move to removing pilots entirely starts (as is happening with cars)? It is, I’m afraid, only a question of time.
1 reply
December 2024
▶ m11
You epitomize the distinction between intelligence and wisdom. Which one is more valuable???
December 2024
The autopilot (however sophisticated and deeply integrated into aircraft systems) doesn’t know how to handle novel situations, can’t weigh multiple possibilities and decide, and doesn’t care if anyone survives.
1 reply
December 2024
Hmmm….
Drifting off topic a bit; philosophy was never really my forté, but if I had to choose one over the other (obviously, both are important) I would say intelligence, based on the logic that if one is not intelligent one is unlikely to eventually also become wise…
December 2024
▶ edfix1
“… doesn’t know how to handle novel situations.” Current autopilots may not (though actually I would imagine they can do this to an extent already) but you have only to spend 10 minutes watching a Youtube video of Tesla’s ‘Full Self-Driving’ V13 at work to understand how far autonomous vehicle control systems have come.
The key question is, are these systems better on average than a human driver. The moment the answer becomes ‘yes’, even if it is only very marginal, lives could start to be saved. I’d actually say FSD V13 is better than the average human driver, already, actually.
People get so hung up on the one-in-a-million ‘Oh, but what if…’ scenario, when in reality it is extremely common day-to-day silly mistakes that human drivers make that cause the vast majority of serious collisions. From what I have seen, FSD version 13 would be preventing at least half of current accidents already. And so, isn’t it logical to accept the one-in-a-million fatality for the sake of 10 or 100 day-to-day ones?
The aviation industry is better than most at applying logic to problems rather than being swayed by emotion (let alone hysteria). For example, the design ‘safety factor’ (ie 'how many times stronger than just ‘in theory’ does a component need to be just to be on the safe side?) for aircraft is very small compared to pretty much any other field of engineering (rockets, maybe?). It has to be because otherwise aircraft would be too heavy to be profitable to operate. If the same safety record of cars was applied to aircraft, the aviation industry would have failed decades ago!
Like I said before, it is only a question of when, not if.
2 replies
December 2024
▶ m11
For many years I managed a lot of programmers and I’m a programmer myself. All this automation in the aircraft is run by sensors feeding data to programs like the ones that my folks wrote. I also took software engineering graduate courses in software safety. Software is great for relieving a pilot of complex or tedious tasks but it remains a tool for the pilot. There is no way for software to react to every possible emergency situation. If a single pilot gets ill or incapacitated (or dies) in flight, then another pilot is needed. Having two sets of eyes and brains to figure out aircraft system problems is better than one set. In addition to the software there’s a network of sensors that feed data to the aircraft computers. These sensors can malfunction and feed bad data to the computers (737 MAX). Pilots are mobile troubleshooters in that situation - something aircraft hardware and software are not. Someday the progress in AI might make up the difference but that’s not today. We all know that having a single point of failure in an aircraft is not a good thing. The same goes for a single pilot in an airliner.
December 2024
Everybody focuses so much on the number of pilots on the flight deck, and whether we go from two to one before we go to none. But I believe there’s another step in the process. The aviator’s equivalent of Work-From-Home. Remote piloting. This is how the military operates their “unmanned” surveillance aircraft. I think there is always a need for a human in the control loop, especially to handle situations that are new or novel or require adaptive thinking. But I am not convinced that there is an absolute need for that human to be physically on the flight deck, provided the datalink connection from the control station to the aircraft is extremely robust. Also, in this way, the humans (yes, most likely plural) in the control station could probably be operating a great number of aircraft, most of which are in stable cruise flight most of the time, and don’t require minute-to-minute attention.
2 replies
December 2024
Flew most of my (airline) life long haul. Classroom style education during ab initio or type conversions don’t come even close to the things I have learned during these flight hours from experienced colleagues, ‘hangar’ talking for hours on end and also keeping both of us awake as an extra benefit. Later transferring my knowledge to the younger generations. That transfer of experience will die with single pilot.
We were allowed to make “NASA” naps, so I have experience flying 4 holers “solo”. But NOT during the crossing of the ITCZ… Or overflying the Andes, Himalayas, or Tibet…(from W to E…).
So to a degree I could live with a few hours alone there, but T/O, departures, approaches and Landing…I see trouble brewing. Listen to R/T tapes any regular day in ORD, LAX or JFK. That’s about a job for ONE guy! Not counting the pilots whose English is not their native tongue.
I agree that, when flying the B747 -2/300 long time ago, I could not fathom we would ever fly this beast with only 2 pilots but the industry did got it done, and when I finally arrived at the LH seat of the -400 I never missed the 3 man’s cockpit.
But this is a complete different case in my view.
December 2024
And therein lies one of the main problems with that concept - the datalink. Unmanned military drones that are used just for surveillance or human-triggered attacks are one thing, but carrying paying customers is another thing entirely. Lose that datalink and there is no human in the loop (and probably at the same time that a human is very much needed to be in the loop).
It’s not just the transceiver on the aircraft, but also the transceiver on the ground and the signal itself. Even if both ends are robust and secure, the signal itself could be jammed (intentionally or unintentionally). And then there’s also the question of how do you secure the link from unauthorized access - armed guards at every remote control facility?
December 2024
I am glad this is not some kind of popularity contest, because I will definitely lose out on this one.
Why would airlines want to reduce from two pilots in the cockpit to one? Do you think it might be related to cost? I will offer the first and painfully obvious solution: airline pilots if they REALLY want to keep the safety of a second body in the cockpit should negotiate compensation plans at 50% or so of their current level. Option #2 is even simpler: Let Cojo be a part of the cabin staff - available at any time needed, At one time every stew had to be a fully qualified nurse, so we have already been there. No reason the top position of flight attendant couldn’t have the level of professional qualification required to use the right seat.
December 2024
I appreciate the humor in the article.
But all joking aside, the flight engineer is a fundamentally different situation. Improvements to computer control and reliability of engines and aircraft systems mean we didn’t need a dedicated person to operate those. And while removing the third crewmember is a slight reduction in redundancy, removing the second eliminates redundancy.
3 replies
December 2024
I knew I liked you. Quoting from the Hitchhiker’s Guide to the Galaxy sealed it for me.
December 2024
▶ bbgun06
Agreed. And to put some math to it, 3 crew to 2 crew = 50% loss of redundancy. 2 crew to 1 crew = 100% loss of redundancy.
December 2024
CrowdStrike. I sure hope it’s not running in the AI that’s controlling my flight. Oh, wait! I don’t fly commercial any more…
December 2024
▶ bbgun06
“Improvements to computer control and reliability of engines and aircraft systems mean we didn’t need a dedicated person to operate those.”
Ask the passengers and crew of Swissair 111 about that. Oh, right – you can’t. They all died when the automation that was used to justify the removal of the FE from the DC-10 when it because the MD-11 was destroyed by the fire, leaving 2-person crew too task-saturated to deal with the cascading problems.
And that still doesn’t address the problem of finding unanticipated holes in the automation, as in the Airbus 320 which refused to follow the pilots’ control inputs at the air show at Toulouse, or the Boeing 767 whose engines refused to spool up because the FADEC’s were programmed to avoid the possibility of engine inlet icing – while descending out of 18,000 west of Denver over the 14,000+ Front Range (fortunately, it got warm enough below 16,000 that the engines obligingly finally responded to the thrust lever commands).
December 2024
Some good points and questionable ones here.
Good point - carrying a plane full of passengers is different that operating souless drones that are mostly clear of civil airways and busy airports.
Good point. Autopilots and automation software is dependent on sensor inputs. Faulty inputs likely result in faulty solution–yes pilots can come up with the same faulty solutions, like Air France and Colgan Air. But how many times do pilots correctly analyze faulty sensor inputs and do the right thing. We don’t know but after 50 years of flying, I’d say it’s not unusual.
Questionable - Tesla’s latest autonomous driving software should make us more comfortable with the idea of turning over the airplane to Hal. I might buy this if every plane in the system was being operated by the same software using the same logic tree to make operational decisions – but then we’d have to ban most of GA from flight.
My opinion - I don’t believe the push to total autonomy is justified given that the current manned system has such an incredibly low Part 121 loss rate - .006 per 100,000 hours or one for every 16.3 million flight hours. I suspect that the creation and maintenance of an autonomous system capable of achieving equivalent results will actually be more expensive than maintaining two pilot crews. I’m also not sure we are capable of doing this as long as we maintain a mix of big and small airplanes using the same facilities. Give pilots all the technology and computed help we can afford but, keep them in the cockpit.
1 reply
December 2024
…no meat puppets in front, then none in the back.
and the AI pilot is getting a check ride before each flight? should only take a few millisecs.
December 2024
Airbus, more so than Boeing, is actively pursuing single-pilot operations, but significant safety concerns from regulators, unions, and passengers make it unlikely to take off anytime soon. Achieving this would require major technological advancements, regulatory changes, public trust—and perhaps even a one-holer fitted with aircraft controls and communication systems to keep things running smoothly.
December 2024
Interesting choice of a movie character reference, a computer operating a ship that killed 1 of the only two crew members not in suspended animation during the voyage. In the 1990’s there was a forum that had the then chairman of GM and Bill Gates of Microsoft. Gates commented how far behind in technology GM autos were at the time. The chairman of GM replied that if their autos had the same software that Microsoft was selling at that time, GM’s autos would break down every mile, waiting to be reboot so they would work again. Ask US Customs how many of their computer controlled unmanned aircraft they have lost at the border. All you have to do is see all the “updates” software companies send out to see just how reliable computers can or cannot be. The amazing thing is that most computer consumers have accepted this as “normal”. Would you want to ride on an airliner that has to have “updates” to software to correct an issue that was not thought of or some error that was originally programmed in? There are some aircraft now that when some system doesn’t work as programmed you have to do a total shutdown of electrical system to reboot that system. Imagine dealing with something like that as a single pilot! I highly doubt in my lifetime we will ever see single pilot airline ops. It would take an act of Congress along with changes in the FARs. Knowing how long it takes for government and the FAA to act, I’m not to concerned about this issue.
1 reply
December 2024
As a gray-haired software engineer, I am both impressed by the progress of technology in my lifetime, but also incredulous how casually we dependent on complexity that most of us don’t understand. I can call my son in Europe with sound quality as good as a local call (but the call drops without warning). My e-assist bicycle exchanges data with my phone better than Apollo and Houston (but flashed red because it needed a firmware update). In both cases, failure modes have minor consequences. I cannot fathom software that adequately handles all the failure modes of a single-pilot, or left-seat-plus-remote-seat airliner.
I can see why check-writers push for automation and business is infatuated with AI. I believe them to be perilously slippery slopes.
December 2024
I’m sorry Dave, I can’t do that.
December 2024
▶ davidbunin
In terms of the current debate (one pilot + automation), the unspoken assumption is the automation will actually be the pilot flying under normal circumstances. This means the human will be the backup, with the argument now being this fallible backup needs backup. Considering this chain of diluted probabilities, I would say a ground-based second pairs of eyes would be adequate even with a less than supernaturally robust data link.
I’m not going to address the social fellowship aspects though…and they aren’t just jokes by any means.
December 2024
HAL also killed ALL of the crew members who were in suspended animation.
December 2024
▶ bbgun06
The 767 brought the flight engineer’s function to the cockpit panel . But Ansett Airlines in Australia bought 767 and the unions would not allow the airline to displace the flight engineers so Boeing was paid to install an F/E panel which duplicated the same functions that became standard on the pilot’s panel. Just ludicrous! After Ansett went bust the receivers had a hard time selling those planes, I read.
December 2024
The 737 Max in Asia is a perfect example of ineptitude and single pilot confusion. The aircraft had an uncommanded pitch up so the captain (who was flying) gave the plane to the f/o (very low time and almost certainly was a seat warmer) to fly the plane while the captain reached for the flight manual. The automation was what downed the plane, and pilot input was inept, and now Airbus is promoting single pilot operation? Admittedly, Airbus started this single pilot talk for freighters, but after gaining experience it would extend to regular flights.
1 reply
December 2024
The determinant factor will be how detached and ignorant the flying public will be to single pilot ops. If insurance rates escalate for single pilot operations then the airlines will follow the money. I would think the airframe builders will consider very carefully their liability as crashes kill hundreds of passengers and people on the ground. Caveat emptor.
December 2024
I think Germanwings proved why we need two on the flight deck at all times.
1 reply
December 2024
▶ Jzarinnia
You tripled the risk of having a nutcase in the cockpit (two pilots plus a flight attendant for bathroom relief).
December 2024
Surely you jest! The 737 MAX grounding charade clearly showed that the “aviation industry” now encompasses the news media, Congress and a number of other similarly incompetent entities who can and do drive those emotional and hysterical decisions you speak of—to the detriment of the aviation industry. From conducting their own investigations into crashes with idiotic results that the official investigations then copy, to pretending aero engineering competence and arbitrarily deciding that a particular airframe shall not support any more derivative models and trying to force such limitation by legislating engineering design, to silly demands for government oversight with goals that simply cannot be achieved, these leeched-on entities and their misguided antics are driving the aviation industry to parity with the car industry.
December 2024
▶ 26981
You got most of that backwards. The 737 MAX experienced a repetitive runaway stab trim, and the Captain handled the resulting uncommanded pitch DOWN each of the 22 times it occurred by reflexively using elevator control and then electric trim-up to assist the elevator. Per design, using electric trim-up temporarily halted the operation of the errant system causing the runaway (MCAS) and kept its operation on hold until the trim switch was released. But the Captain didn’t seem to know what to do next (apparently had not been trained on the runaway trim procedure), but the FO—who wasn’t low time, had ~5000 hours in the 737, but was deficient on just about every proficiency check throughout his career—couldn’t find procedures in the manual, so the Captain handed control to the FO so HE could dig through the manual to find a procedure to address the runaway. The FO didn’t seem to know how to use the trim controls to assist with pitch control and thus failed to counter the runaway, which put the airplane into an unrecoverable dive…while the Captain had his head buried in a manual. So the effectively single-pilot Captain wasn’t confused, and automation didn’t down the airplane; lack of pilot training combined with the FO’s gross incompetence did.
December 2024
With the number of airplanes zipping about in the skies these days, how many calendar hours does it take for the world fleet to fly 16.3 million hours?