It definitely is not useful. Your model should at least attempt to approximate reality, not to depart from it by putting effect before the cause. That way lies madness.

It is not a model. It is a description. I'm torn on whether it would be correct to refer to the approach as constituting a sort of analogy.

No idea why you think the effect is being put before the cause. I'm hungry so I head to the kitchen. An observer says "he wants to eat". Antibiotics are administered. Only the bacterial cells expressing a certain set of proteins survive. An observer says "the infection wants to be resistant".