This remembers me old history of fail redundant space computers, which appearing first lead to less failures, but in some short time things got much worse, beca

This remembers me old history of fail redundant space computers, which appearing first lead to less failures, but in some short time things got much worse, because people give up on software reliability. Most expensive documented example - failed first launch of Ariane-5, where it's 3 fail redundant computers all stopped to work on same exception - costs 1 billion Euros. Used inherited software from Ariane-4, which is just have too much different design, so in some flight modes Ariane-5 could achieve much higher speeds than Ariane-4. And over time these things where written in Ariane-4 software as constraints and trigger exception if reached. So it was not the question, how redundant where computers and how many computers working in parallel as hot replacement - even if where 100s, all will become non working.