Part of the "benefit" is the software breaking again in the future. Like I said, there is no cost to software failing and even a benefit. If people died when software failed, that 25x effort would probably become 5x dude to efficiencies of people actually working on the problem