Once you fully internalize that your buggy program is running on a buggy framework, in a buggy language, in a buggy sandbox, on a buggy virtualization, on a buggy file system, scraping along on buggy silicon running buggy microcode, managed by other buggy silicon running buggy firmware, using peripherals with their own buggy silicon and firmwares, with everything happening billions of times per second (a car engine typically doesn’t reach a billion cycles over a 20 year lifespan ), it seems mind bogglingly improbable that anything works at all lol.
The fact that it does work, and can even routinely reach 5 nines, is a testament to the generosity of the universe… and a good argument for making sure that to whatever extent possible, you should write your software to be resistant to random events and erroneous operations. Fail safe/ fail soft, fail and retry, fail and restart. The more resilience we build into our work, the less angst we create in the world.
It’s amazing to see both the half hour miracle of engineering and the half century one playing out at the same time. I just hope our accomplishments today will in some way be as durable as those of generations past.