I kind of get the worry, but the requirements and processes seem to scale exponentially with reliability needs. Online companies may fuck up every day in new and creative ways and we barely get to hear about it. On the other hand we know of every nuclear failure so far with enough public details to discuss the whole time line, system design, steps each person followed, etc. and the death count is still minimal. Then each of those is an input to the future processes. Nuclear power plants and air traffic are in their own class of reliability and safety processes - not even comparable to that's happening in internet commerce.

The Italian cable car was really messed up. The emergency brakes of that cart were intermittently triggering, so the operator jammed a piece of metal to stop that from happening.

His assumption is surely, "Relax, what's going to happen, the cable won't break!".

There are also plenty of services with really incredible uptime. You just don't hear about them because they're deep in whatever stack you're using and haven't broken publicly in decades.

It's all about good engineering practice and architecture.

In fairness, the government does not watch over your shoulder if you write code for payment apps. Nuclear energy oversight is so over-the-top, it's almost not worth doing it.

Yes! We will always make huge mistakes.

.. so we need to accept the eventuality that the worst result will eventually occur.