Off-topic, but I guess this moved down a spot before the second page loaded so it showed up twice. Gave me a chuckle considering the title. https://imgur.com/a/HMyuDkl
The failure modes of software and hardware are different, and this needs to be taken into consideration.
If you take a reel of 1000 resistors, and build a bunch of boards with them, then one fails: local redundancy can help, and the worst case is the failure of the board around that resistor.
If you take something like left-pad, then each usage of it is not a separate component, they're all the same component, and will fail globally all at once or not at all.
Most software fails differently: in 95% cases it works 100% of the time, in 5% cases it always fails. You can't use two identical pieces of software for redundancy.
So this kind of redundancy only makes sense in rare cases:
1. There are two different pieces of software that do the same thing.
2. They are interchangeable enough, you don't need a fat adapter for each one.
3. The fault is not in one of their shared dependencies.
1 and 3 are rare because we try to reuse as much software as we can.
That said, this does happen a lot with web standards: if a standard exists as an upgrade over a previous standard, being able to fall back to the old standard is a common requirement.
