4 ms·

Two new potentially fatal bugs found in 737 Max software

36 points by rpiguy 7y ago

rpiguy 7y ago
This article downplays the magnitude. Loss of vertical stabilizer control is very, very bad...
contingencies 7y ago
Obviously they are not writing this stuff with provability in mind. Obviously emergent state occurs with numerous subsystems however it's not like the same issue isn't shared in other complex systems... culture of technical debt and commercial shortcuts must have overtaken a culture of testing for a very long time to let this happen. People were raising issues years before the latest 737 Max issues: Oh yeah. I'm not flying on a 787 [...] I would definitely avoid flying on a 787. - Cynthia Cole, former 32 year Boeing employee and former President, Boeing Engineers' Union ... via https://github.com/globalcitizen/taoup https://github.com/globalcitizen/taoup Amusing(?) idea for outlandish conspiracy theory (in the spirit of 'virus all the things!'): bad actor within US bioweapon agency goes rogue, misrepresenting government as effective Boeing stockholder and major client releases COVID-19 to slow travel sector and mitigate impact on company and distribute pain and suffering to Airbus.
- rpiguy 7y ago
  The 787 has to be rebooted every 51 days or it shows faulty data to pilots: https://www-theregister-co-uk.cdn.ampproject.org/c/s/www.theregister.co.uk/AMP/2020/04/02/boeing_787_power_cycle_51_days_stale_data/ https://www-theregister-co-uk.cdn.ampproject.org/c/s/www.the...
markrages 7y ago
> Hopefully, these are the last software flaws. I have tried this approach and can report that "hope" doesn't accomplish this.
- shaftway 7y ago
  Our reliability engineers use "Hope is not a strategy" as one of their branding statements.
  - contingencies 7y ago
    Nice one. Added to https://github.com/globalcitizen/taoup https://github.com/globalcitizen/taoup
LatteLazy 7y ago
It's almost as if you shouldn't fix aerodynamics problems with software...
- qchris 7y ago
  I believe that your point stands in the context of a large-capacity commercial airliner, but it is worth considering that in some fighters, this is actually by design. For example, the F-16 is not passively stable--even when the pilot lets go of the stick, the control surfaces are still working to keep the flight moving properly. However, there's much lower benefit when, you know, you're not designing a plane to dogfight with other planes in a life-or-death struggle thousands of feet in the air. Even then, I believe that the control systems are based on voting, where many operations occur in triplicate between various processors and then a consensus is reached for the action. But building a system like that is difficult, unnecessary for a craft like the 737 MAX, and crucially didn't come close to meeting the same level of redundancy, since I believe the crashes were traced back to single-point failures in the angle-of-attack sensor. Boeing tried to have it both ways, and ended up killing hundreds of people as a result. Anyway, I thought it was worth pointing out that sometimes fixing aerodynamics problems with software is actually used in airplane design to good effect. The problem is that no one should have thought about or signed off on doing so in this situation.
  - contingencies 7y ago
    The problem is that no one should have thought about or signed off on doing so in this situation. Perhaps a broader problem is that a culture was created that facilitated that.
  - aaronmdjones 7y ago
    > Even then, I believe that the control systems are based on voting, where many operations occur in triplicate between various processors and then a consensus is reached for the action. > But building a system like that is difficult, unnecessary for a craft like the 737 MAX Why? The 777 has it. http://www.citemaster.net/get/3c501aaa-39d5-11e4-9cb6-00163e009cc7/R8.pdf http://www.citemaster.net/get/3c501aaa-39d5-11e4-9cb6-00163e...

36 points by rpiguy 7y ago

This article downplays the magnitude. Loss of vertical stabilizer control is very, very bad...

Obviously they are not writing this stuff with provability in mind. Obviously emergent state occurs with numerous subsystems however it's not like the same issue isn't shared in other complex systems... culture of technical debt and commercial shortcuts must have overtaken a culture of testing for a very long time to let this happen. People were raising issues years before the latest 737 Max issues: Oh yeah. I'm not flying on a 787 [...] I would definitely avoid flying on a 787. - Cynthia Cole, former 32 year Boeing employee and former President, Boeing Engineers' Union ... via https://github.com/globalcitizen/taoup https://github.com/globalcitizen/taoup Amusing(?) idea for outlandish conspiracy theory (in the spirit of 'virus all the things!'): bad actor within US bioweapon agency goes rogue, misrepresenting government as effective Boeing stockholder and major client releases COVID-19 to slow travel sector and mitigate impact on company and distribute pain and suffering to Airbus.

rpiguy 7y ago
The 787 has to be rebooted every 51 days or it shows faulty data to pilots: https://www-theregister-co-uk.cdn.ampproject.org/c/s/www.theregister.co.uk/AMP/2020/04/02/boeing_787_power_cycle_51_days_stale_data/ https://www-theregister-co-uk.cdn.ampproject.org/c/s/www.the...

markrages 7y ago

> Hopefully, these are the last software flaws. I have tried this approach and can report that "hope" doesn't accomplish this.

shaftway 7y ago
Our reliability engineers use "Hope is not a strategy" as one of their branding statements.
- contingencies 7y ago
  Nice one. Added to https://github.com/globalcitizen/taoup https://github.com/globalcitizen/taoup

LatteLazy 7y ago

It's almost as if you shouldn't fix aerodynamics problems with software...

qchris 7y ago
I believe that your point stands in the context of a large-capacity commercial airliner, but it is worth considering that in some fighters, this is actually by design. For example, the F-16 is not passively stable--even when the pilot lets go of the stick, the control surfaces are still working to keep the flight moving properly. However, there's much lower benefit when, you know, you're not designing a plane to dogfight with other planes in a life-or-death struggle thousands of feet in the air. Even then, I believe that the control systems are based on voting, where many operations occur in triplicate between various processors and then a consensus is reached for the action. But building a system like that is difficult, unnecessary for a craft like the 737 MAX, and crucially didn't come close to meeting the same level of redundancy, since I believe the crashes were traced back to single-point failures in the angle-of-attack sensor. Boeing tried to have it both ways, and ended up killing hundreds of people as a result. Anyway, I thought it was worth pointing out that sometimes fixing aerodynamics problems with software is actually used in airplane design to good effect. The problem is that no one should have thought about or signed off on doing so in this situation.
- contingencies 7y ago
  The problem is that no one should have thought about or signed off on doing so in this situation. Perhaps a broader problem is that a culture was created that facilitated that.
- aaronmdjones 7y ago
  > Even then, I believe that the control systems are based on voting, where many operations occur in triplicate between various processors and then a consensus is reached for the action. > But building a system like that is difficult, unnecessary for a craft like the 737 MAX Why? The 777 has it. http://www.citemaster.net/get/3c501aaa-39d5-11e4-9cb6-00163e009cc7/R8.pdf http://www.citemaster.net/get/3c501aaa-39d5-11e4-9cb6-00163e...