C is Brittle D is Plastic

[Richard (Rikki) Andrew Cattermole]

On 09/04/2026 9:33 AM, Walter Bright wrote:
> On 4/8/2026 1:48 PM, Richard (Rikki) Andrew Cattermole wrote:
>> Given how prevalent software is now in aerospace, the fact that planes 
>> aren't falling out of the sky on a regular basis is pretty incredible. 
>> No amount of hardware can make up for how invasive it is.
> 
> Having done design work for 3 years at Boeing, I can tell you that the 
> software does fail. So does the hardware. And so does the nut behind the 
> wheel.
> 
> The reason planes don't fall out of the sky is:
> 
> * backup systems and workarounds *
> 
> (I don't know how often this happens, but airplanes are given the green 
> light to fly even when many things are broken. There is a "minimum 
> equipment list" which specifies what cannot be let slide.)
> 
> The reality is one cannot make perfect parts. But one can greatly reduce 
> the consequences by making the system *tolerant* of faults. It's a 
> different mindset.
> 
> Note that the Fukushima reactor and the Deepwater Horizon rig did not 
> have backup systems. And so when something went wrong, a zipper effect 
> resulted.
> 
> (I don't recall the details, but I went through the zipper for both of 
> them. It was rather astonishing for me as redundancy and fault tolerance 
> was hammered into me.)

Over the last 50 years, software has replaced hardware, and simpler 
hardware became more complex. Both introduce their own new risks that 
didn't exist when you were working on it.

The significantly increased surface area and responsibility of behavior 
controlled by software, and the fact that aircraft are still flying 
without catastrophic failures on a regular basis due to software alone 
is what is so amazing.

While I'm sure we could if we really tried, could find examples where 
software failed and could not recover, finding hardware faults causing 
software faults is a lot easier than just software faults by itself.

I.e. 
https://www.eplaneai.com/news/airbus-to-update-a380-engine-software-by-q1-2026

Note: I picked Airbus to search for, because I know that they use Astrée.

By all accounts, we should be seeing a lot more catastrophic level 
problems with articles on software failures in aircraft, yet we are not. 
The surface area is simply too large for humans to be the only ones 
catching problems.