An argument against the present utility of trying to do brain emulations is that we cannot predict or simulate the behavior of Caenorhabditis elegans, despite having a complete connectome map of its simple nervous system. So why do we think replicating the much more complex human connectome computationally would be any more useful? These criticisms have not stopped some groups from forging ahead. Theodore Wong and team at IBM has now published preliminaries on the most ambitious brain simulation to date, with 10^14 synapses.
Certainly simulating a human nervous system is the end goal of all this, but it seems a lot of money and work is going toward projects when there is a very glaring failure at an earlier level of the same enterprise. As long as your project is being funded by a large institution with deep pockets, you probably don't have to address this question; at least not yet. But if a counterargument exists to "simulating C. elegans doesn't give us anything useful so simulating H. sapiens is also unlikely to do so", I haven't yet seen it.