The point is that in real life problems it's more important to throw away ridiculous "assume that"s than having very neat models.