True, but I think his point was more precisely that when you go long periods of time _without_ having those systems fail (hopefully in relatively small ways), you get trouble. If Visa/Mastercard had outages on a regular basis, no doubt our retail system would be quite robust to them...

not sure if this counts as a paradox but that systems that are more reliable cause more chaos when they go down . The person easiest to survive the apocalypse is someone who already lives it.

I believe that is why Netflix has the "Chaos Monkey" -- to occasionally break things and see what happens.

https://medium.com/netflix-techblog/the-netflix-simian-army-...

Yes! That is a brilliant point. You have to test for failures, it's not obvious - if you have failing systems occasionally you are already testing for failures.