well, I'd say at this point the science behind soundstage is decently well understood. It's HRTF + angle the sound hits your ear + area of the ear it impacts + micro adjustments your brain makes when your head moves tiny amounts.
The top headphones have gotten down angling (why most of the top headphones use some method or another of angling the drivers). Also most top headphones these days produce a decently wide sound array that allows the sound to hit the folds of your ears, through a combination of giving the driver a bit of space from your ear and large drivers with fewer "hotspots" on the driver. This is largely where we've seen the great expansion of soundstage in the last 10 or so years, as makers understood and implemented these two ideas. HRTF has been more frequently and better integrated into amps and DACs in recent years as well. the micro adjustments thing is probably a couple years off, as VR technology will eventually allow a sensor to monitor your head position and digitally process the HRTF algorithm before sending it to your headphones. At which point we can potentially have real, genuinely immersive soundstage.
Getting back to the current level of tech:
My take on what is "natural" vs what is "false" or exaggerated" when it comes to soundstage is how it reproduces the intended size. With most studio created music this isn't even possible to suss out because there isn't a clear frame of reference, and unless you were in the control room, talking to the producer and engineer when the record was made, you really don't even have a clear idea of what they were after. However, with live recordings, you can often get a feel for how it should sound, especially if you have experienced the venue being recorded.
My take on the HD800 is that it has the widest soundstage that doesn't sound preposterously exaggerated using gimmicks. and it has decent/very good depth. But, like all headphones that haven't been subjected to some form of crossfeed, it can suffer from being a bit too far on the left and right of you, with dropouts in the middle left and middle right, the infamous headphone "disconnected" soundstage; a blob of sound in the center, a blob of sound on the right and a blob of sound on the left, which noticeably gets less "dense" between those areas. That's simply because most music is still engineered for speakers, not headphones. On music that was binaurally recorded for headphones in a proper concert hall, the HD800 can sound stunningly lifelike, almost approaching a great pair of speakers in its ability to reproduce lifelike sounds.
The other pro is that since the HD800 goes a little overboard in soundstage width, when "corrected" with a good crossfeed implementation (be it DSP or analog circuit) it becomes more "connected" and a bit deeper in soundstage, without becoming too narrow. With many headphones, if crossfeed is implemented, the soundstage "connects", but is then in turn too narrow, So you have to choose between lifelike width, or lifelike connectedness. I personally choose connectedness, as nothign drives me crazier than the hard left and hard right sounds that don't even seem to be coming from the same space, like you're listening to different rooms, but it would be nice to have that, without everything sounding pushed towards the center, and few headphones cna do both, even with DSP. With the HD800 and a good crossfeed implementation you get the best of both worlds. At that point, the only thing missing from the soundstage recipe is the ability to tell when your head is moving and micro adjust the auditory image based on your headphones, and that's the next big leap for headphones down the pipe.
So, yes, I agree that the HD800's soundstage, when uncorrected, is "unnatural" but so is every headphone on the market. The pro of the HD800 is that you can fix it with crossfeed, whereas most headphones, with crossfeed, the soundstage sounds more natural in some ways, but can also sound claustrophobic or at least "pushed towards the center". I'll be excited to see how the Elear and Utopia fare in this regard.
edit: also, it's perfectly fine to like an overly large soundstage, just like it's perfectly fine to like a headphone that has more bass than is really neutral, or more treble. This isn't a comment at all on what is "best" but rather, what gets closer to reproducing the sound intended by the creators. It's a perfectly legitimate thing to say "screw what they intended, I want this to sound even wider." And a lot of people certainly make that choice in the headphone world. Some people chose headphones for convenience, or cost, but some people want more resolution and more extreme left and right sounds than music makers ever intended, and that's why they chose headphones in the first place, over speakers.
Exhaustive explanation on what soundstage headphones setup is. Speaking like a sound engineer.
I know of sound engineers who would comment like you but instead of the HD800 they EQ and monitor with the LCD-X.
My purpose is to get the best soundstage (through non binaural registering like you point it as being the panacea) with NON EQ headphones. Actually quite near the goal with the LCD4.
Wondering about the Utopia....