Large systems whose normal operation relies on the availability of one or a small number of individual components are by definition less reliable, less robust to attacks, and less resilient to disasters than large distributed systems which don't have such reliance and therefore operate smoothly in the face of routine failure of a significant portion of their individual components.

This is true for geographic networks, economic and financial systems, software-as-a-service applications, and even potato variety in farming systems.[a]

[a] https://en.wikipedia.org/wiki/Great_Famine_(Ireland)#Potato_...