Aren’t you kind of answering your own question here? If we’re doing things you aren’t (close-in turns, complex routing off simultaneous parallels with minimum separation), why is it a surprise we would have different safety protocols than you do?
Your argument seems to have evolved into “this wouldn’t make sense in Europe so it doesn’t make sense anywhere.” Surely if you’re involved in any way in aviation human factors, you’re aware that there are various regional operational differences between Europe, Asia, the Middle East, North and South America etc., that are designed to account for their specific environments?
I just don’t understand how this wouldn’t make sense to you.
Here are the Attention All User pages for the RNAV departures out of KATL. Not all airports have these; KLAX doesn’t for instance, but KATL was one of the test airports for these things. Items 3 and 4 give you an idea of the level of concern over a minor error.
Because this does not exist in this part of the world. A heading does not happen here in any kind of RNAV procedures, we always fly tracks. Where the two track intercept (095 with 115 GRITZ) a RNAV waypoint would be created, in this case as a fly-by waypoint.
Departure runways are never assigned during taxi, apart from the occasional runway change, but even those hardly ever happen during taxi. With those kind of procedures I can understand the extra mitigations. I’m pretty convinced now we are not overthinking anything here .
Headings or tracks can be coded into a procedure. I suspect (but I’m not certain here) that the reason for the initial heading to intercept a track is that not all aircraft can select LNAV on the ground. If it can’t be selected until airborne (usually after initial flap retraction), the flight cannot technically be cleared to fly a track off the ground, it needs a heading until LNAV is engaged.
Yes. Hence the determination that a little extra safety layer wasn’t a bad idea.
Yes. Yes, I think it is. Funny thing is, I’m pretty sure I’ve actually flown this departure now that I see it, because I seem to remember us discussing how it didn’t make sense they were using the RNAV departures procedures on it.
Man, what happened to the good old days of “runway heading, cleared for takeoff, shoreline below 2000?”
Oh KATL can be a right pain in the ***. You want non-standard phraseology? Taxiway D at that airport is called “Dixie”, and it’s a major, central taxiway. But god help you if you call it delta, because that might cause confusion with the large airline of the same name hubbed there… and also because the American deep south has a special love of the word “dixie”.
It is a valid path terminator, but not giving accurate and repeatable ground tracks (for obvious reasons). I used to fly the old ATR -500 series, we weren’t allowed to select LNAV until flaps retracted and at or above high bank speed (because the LNAV only operates in high bank mode). We could and were allowed to fly track to fix during initial departure raw data. The Embraer E-jets had the same problem during single engine operation on the old software loads.
I think I have even flown this one in real life. Its not very common, funny enough track to DME distance is not considered as a RNAV path terminator. Headings in RNAV procedures are very uncommon, if they exist at all here. Reminds me how :■■■■: these Jeppesen plates are, they are so cluttered compared to Navblue (Navtech) or Lido.
.
Funny thing is, I’m pretty sure I’ve actually flown this departure now that I see it, because I seem to remember us discussing how it didn’t make sense they were using the RNAV departures procedures on it. 
