Indeed we might never be able to measure the global normalization of the likelihood function of life, but that doesn't mean we can't investigate it's local properties. In fact, I think we can learn a lot from doing just that.