>“Only one computer was used in the past, because Boeing was able to prove statistically that its system was reliable, the person said.”
Years back, my father co-wrote a paper on the Space Shuttle Challenger showing issues with the statistical thinking at NASA:
http://www.math.montana.edu/shancock/courses/stat401/Dalal_e...
I wonder what statistical techniques Boeing used, and how defensible those techniques were.
I can understand a backup if the first fails, but why two identical systems contemporaneously computing?
But if there's a human in the loop and a manual alternate control pathway, detecting a disagreement allows you to cue the manual operator and transfer control to them. Or fall back to a much simpler system of computer aid.
With 1, hardware failures are extremely hard to detect at all, as even your computational checks for internal consistency are subject to mutation.
Unless all 3 different give different results, two failures and one correct.
IIRC the shuttle had a 3+1 system 3 as a cohort with voting and if they couldn't reach consensus the 1 was a minimal system that could keep the lights on.
They said it's common now in critical systems to use three computers and ignore any single computer that disagrees with the other two.
Boeing probably has a similar thing for the fly by wire fighter jets they are involved in but there passenger planes are still mainly directly controlled by the pilot.
Some material: https://www.researchgate.net/publication/26587285_Challenges... https://www.researchgate.net/publication/220845884_Approache...
I spent several years doing stats/BI in the medical manufacturing industry and have on many occasions done analyses that showed what my employer wanted and would stand up to scrutiny, but if my employer wanted the opposite outcome of that analysis, that would also be possible and also stand up to scrutiny.
The best way to address this is to remove any financial/business incentives form the analysis, but that's not really feasible most of the time and also requires larger/more expensive teams that are capable of internally challenging their own work.
https://en.wikipedia.org/wiki/Lies,_damned_lies,_and_statist...
Despite the title, it's mostly an essay on why visualization and presentation are critical to any statistical treatment.
I think it's one of those difficult skill combinations. You want someone extremely well-versed in statistics, plus someone capable of giving a damn about accurate presentation, plus someone with the UX intuition to make the correct choices to make something readable without sacrificing accuracy and important nuances. And if that "someone" is actually two people, you run the risk of telephone misunderstandings as work passes between them.
Which is a tall order.
Made taller by the (as you note) different kinds of audiences: simply ignorant, aggressively biased, rushed, etc.
I suspect they're using that sense of the word "proven", as opposed to a formal software proof.