By the way, I’ve been tormented by this question since visiting NASA – does it make sense to use a multilayered complex architecture, like the whole Windows with drivers, for solving algorithmically simple tasks, which include both moon landing and ballistic missile control? After all, a simple algorithm directly controlling a microcontroller can realistically be tested for all branches, whereas in a complex system one has to introduce the concept of faith and hope. I mean, if a manufacturer promises something, and it’s lengthy and complicated to verify, it’s easier to believe that the manufacturer has already considered everything and to address problems as they arise, hoping that there will be no problems at all. The result of your work goes as a component into a more complex system, where they now trust you. And so a cascade accumulates, at the end of which unreliability arises.
Hi, I'm Rauf Aliev. 