writes:

How many single points of failure in a vacuum system?

Two or three, I suppose. Does every instrument in the cockpit depend on
vacuum?

How do you know how much testing and of what type?

"Too much complexity"? What in Sam Hill does that mean?

I know that the number of possible paths in complex software expands
exponentially with complexity, and that it is impossible to test them all.
Therefore many paths go untested, and if a bug should exist on an untested
path, and if that path should be executed in operation ... on that path lies
danger.

Even NASA missed things occasionally in testing, and Garmin doesn't have the
time, the budget, or the mindset of NASA. And the computers NASA used were
orders of magnitude simpler than what Garmin is using.