Survivor bias doesn’t mean that everything is a fluke. It just means that some things are flukes and it’s important to include that in our models so we don’t think something random is non-random.

This is very different than everything being 100% random.