Same here. Whether P25 on our state system, EDACS ProVoice (on real PV radios, no less) or D-Star on amateur radio there are still times where the signal becomes garbled / unreadable. The sad thing is, in many cases I think a regular FM signal would still be understandable.
From my experience using D-Star, I think the pundits who push the "hear it or don't" line are basing that on a very specific test setup where the two ends are in fixed positions. That's how I first heard D-Star, as well as most of the demos online, and yes the difference between FM and digital for weak but steady signals is quite impressive.
However, have one end (or worse - BOTH) be moving at the time and you'll frequently have deep nulls that only last a short time. Either from obstructions, reflections / multipath, whatever. With FM we just hear a brief dropout or white noise that is easily overcome but with digital it can cause the decoder to lose enough data to become unintelligible garble for a fraction of a second - or worse, lose sync and take a second or two (or longer) to sync up with the data signal. (Unfortunately, when this happens on D-Star - or at least Icom's implementation - the vocoder doesn't silence the output and the listener is treated to a cacophony of bleeps and electronic sounds. Thus someone is said to be "R2D2ing again"...)
I like digital, but only in applications where it makes sense. If I want to be able to route the traffic elsewhere easily, it's a great fit. But I don't see it being very useful in the situation described, and just adds that much more complexity to the radio. (Well, I suppose it IS useful to the radio manufacturer that "expert" probably works for - they get to sell more expensive gear! -- Yeah, I'm cynical...)
