If you want to pm me some info on why HRTF applies when the driver is close to the ear and not for real life or speakers, that would be cool. Also would like to know if a sealed chamber around the ear creates unique hearing in all of us, then whether velour negates that since it doesn't seal. And whether the K1000 has less HRTF issues.
Actually, HRTF applies to speakers as well - but it's the complete HRTF, which is completely compensated by the brain. Headphones remove at least the torso part of it, even K1000 and open ones. (K1000 allows partial reproduction of head shadowing, but still not torso response.) IEMs remove the auricular part and depending on insertion depth, change ear canal resonance (acoustic impedance).
This means that brain will overcompensate some parts of the sound. So what we're actually dealing with is applying
inverse partial HRTF.
HRTFs can change over time somewhat, as they are based though on physiology.
Sealed ear canal creates a resonant mode depending on ear canal shape - normally ear canal is not sealed on either end, having modes similar to "open pipe", closing it converts them to "half-pipe", changing resonant frequecy by half or so, depending on the extra chamber size added our of the ear. It's as simple as that.
I'm very open to quantifying the variability in human hearing. This could be done, with error bars, clustering, etc. Also quantifying fit-refit variability and measurement variability. (It seems purrin has partially done the latter - I'd prefer more rigorous presentation of that info.)
If we get that information, we can devise a headphone and/or IEM that sounds flat for most people - truly neutral.
Obviously there are other pitfalls, like the namesake, effin' ringin' and other kinds of distortion.