This is exactly what confuses me -- we're modeling both the inputs and outputs, and not with particularly robust validation or verification. Where is the utili

This is exactly what confuses me -- we're modeling both the inputs and outputs, and not with particularly robust validation or verification. Where is the utility? We've established that things are changing. Sure. But if you gave someone a 'stock predictor' of similar quality, few people would choose to trust it with money. But in reality, we dedicate billions of dollars of resources collectively to the outputs of biome modeling.