Fault-tolerant design
Let me take you on a detour through aerospace systems design: when we have to engineer spacecraft that will operate for twenty years in geostationary orbit without any maintenance (imagine your washing machine running 24/7 for 20 years without interruption), or when we have to design aircraft that will fly a billion flight hours without an accident (imagine your computer running for 117,000 years without hanging), we need fault-tolerance and robustness. The way we achieve this robustness is through redundancy and diversity. Whenever a system fails, there is a similar backup system that can take over (redundancy). Sometimes, even the redundant system fails eventually. The most frequent case is because there was a common design flaw in both the nominal and redundant systems. When we want to ensure robustness even against such unforeseeable events (we never know where our design will be flawed, but we always know it can be), we design another backup system that is completely dissimilar from the nominal system (diversity). Generally the backup system design is much less efficient, but much more rugged, therefore it has a lot more tolerance to off-nominal situations. This will allow the pilot to land the aircraft or give ground operators time to reconfigure the spacecraft. The more diversity we build into the system, the more robust the design will be.
Biodiversity theorem
I recently read an interesting analysis in a gardening book: the author was showing that there were always ‘magic’ weeds which showed up after an ecosystem had been upset, so that it always seems to be just the right weeds to improve the soil and make it ever more fertile. [more]