Unless they've come up with a modulation scheme which overcomes the multipath distortion of transmitted symbols, I believe line of sight is practically a necessity.
Hmmm, I don't think so -- at least not in the strict sense used for microwave. A proper line of sight would certainly reduce both attenuation and multi-path. Both are a problem at microwave frequencies. Which is why n*GHz radios won't sell many units to the average ham at home.
But UHF has the advantage of not being hampered so much by attenuation from trees and the like. That's why it's attractive for us, even if the bandwidth is lower. And, UHF multi-path isn't as much of a problem as for GHz, but still more of a problem than at VHF. So it will work will without clear line of sight, but not from all the places that VHF packet would work. FEC will be essential.
As for new modulation schemes: I believe NWD was going to be using a chipset from a commercial data radio (I don't know the details). The "not reinventing the wheel" approach sounded good to us, because we pride ourselves in our uptime/availability for EmComm use. But one would also hope that the radio would be capable of software upgrades to modulation schemes as well. That would be attractive to those who are into experimentation.
Bottom line: Many projects fail because they focus on speed or some other esoteric means to an end, but not on the end itself -- the application. And without an application, there are no users. And without lots of users, the idea will die. So we don't look at a potential 56k-100k UHF radio as much slower than microwave and narrower coverage than VHF packet. We look at it as way to carry what is probably the most ubiquitous application in the world today: e-mail. And, by the way, it happens to have MUCH, MUCH broader coverage than microwave (the ONLY possible coverage for many), AND be MUCH, MUCH faster and more feature rich than packet.
Michael N6MEF