You seemed to be saying that the Harman curve (black line) is what a neutral headphone should follow. This implies that the HD 800 and SR-009 have horrible measurements and are not what people say they are -- neutral reference headphones. If this was not your implication, it is up to you to clarify.
Ok, first of all, I wouldn't be ranking the HD800 at the top of my most accurate list if I didn't think it was a great headphone. Its treble is elevated above a neutral response for what I'd consider an accurate headphone, so mods and/or EQ is necessary. Never in my posts did I say or even imply that the HD800 (or even the 009) were horrible measuring headphones. My giving understanding to how the target response curve was arrived at has nothing to do with saying X or Y headphone measures poorly.
Second, on the notion of measurements, I've made a point of saying that FR measurements aren't the end-al tell-tale picture of how a headphone sounds. You still need to take into account the time domain, distortion and other qualities of the sound that objective data can not tell you. There are lots of amps that measure with perfect frequency response and distortion numbers well below the audible threshold, but they all have very subtle differences in sound. The same is true for headphones. As I mentioned previously, a headphone with a notably boosted low-end on a FR chart doesn't have to be more bassy than one with linear measuring bass, like the LCD-X vs FSP example.
The harman preferred listening curve has a lot to do with developing a new target response for headphones. You can go back and reread some of Tyll's articles on the matter, but it essentially amounts to emulating a pair of flat measuring speakers in a good room. The green line in those pictures is a pair of flat measuring speakers in an anechoic chamber measured in a good room. There will be some natural treble attenuation from the room itself. Headphones are placed right by the ears, and could possibly need more attenuation in the treble, and since headphones only play music into the ears, there's no gain from the bass taken in by the human body itself. Anybody could tell you that a speaker system measuring flat will have way more powerful bass that's both heard and felt than just a pair of headphones. I think you're underestimating just how much your body plays a role in shaping the overall sound that you hear.
Now back to the HD800. It actually does have a small bit of mid-upper bass emphasis over its midrange, while the upper-midrange is slightly recessed. I consider this a very realistic sound. The only area in which the HD800 is lacking is 60hz and below, where it could use 3-4db or so more bass. For what it's worth, I correct that using EQ. Taken as a whole, the HD800 is a brighter than neutral headphone, yes, but I'm not turning a cold shoulder to it and labeling it as a bad measuring headphone just because it doesn't exactly follow the black line. Overall, the listener preferred curve on the Harman HRTF will help shape an eventual ideal headphone response curve, but a headphone's FR will not always tell you if it's the most accurate or not. Do note how I said in my first post of the topic that the PM3 was the closest to the preferred listening curve, but that it wasn't the most accurate sounding. My saying that it wasn't the most accurate sounding has nothing to do with me possibly thinking that it's too bass heavy, because it 's not. Some headphones need the extra bit of boost, while some do not.
TLDR, stop showing us these dumb measurement pictures.