You're accepting his defeatist attitude too quickly. The problems he lists are easily surmountable. The military uses mesh networking to a great degree of success. A globally-distributed up-to-date copy of the network topology is a faulty starting point.
So is something like Multipeer Connectivity Framework in existence mostly because it is a much weaker version of true mesh networking? (Or not mesh networking at all?)
Do you think weaker versions of mesh networking that don't solve the underlying problems could also be valuable?