Quite right. Also, trying to prevent harm through prediction is vastly more expensive than trying a product out in the market, and gets more expensive the more

Quite right. Also, trying to prevent harm through prediction is vastly more expensive than trying a product out in the market, and gets more expensive the more harm you try to prevent. It also causes product development paralysis, as deeper predictive scenario analysis results in longer design times and design lock-down ahead of product release, reducing agility. While we might want all products and designs to be safe, in reality, we can only economically do so much before the cost of predictably assuring safety becomes exponential. See NASA.