@svart-hvitt , heres the thing about the Harman headphone curve, which inherits the same uncertainty that plagues in room loudspeaker targets. We dont know how to objectively eq a signal for different room contributions (the combined effect of on and off axis radiation, and room contribution) to arrive at a true neutral response. Not even for a given room and DI, never mind different rooms and DIs. In fact, we know its not even quite possible.
So Harman wisely put some stakes in the ground: first try to define what is a neutral speaker in a room. There is no way to DBT this because you dont have access to the original performance space and because by necessity the best (neutral) recordings are artistic attempts, not purely objective, interpretations. So, using DBT, they arrived at targets for speaker design that have consistently shown best outcomes for trained listeners attempting to allocate highest score towards perceived neutrality (one has to appreciate the lack of anchor and the limitations of auditory memory), testing various responses and DIs. They did their best to study if the outcome varied by gender, age, experience, culture, and rooms. The outcome held across these variables.
Based on this, they had a pretty solid theory on what statistically a neutral speaker measured like.
They then took the best available representation of this speaker and placed it in a room meant to be representative of an average listening room. I believe its an IEC room. I've commissioned an IEC room, had it built, and tested its acoustics, so I'm satisfied for myself that its a reasonable choice. They then treated it so it didnt have any gross acoustic anomolies.
As I understand it, they then used a dummy head to record the in room response of this speaker in a most neutral placement. Different dummy heads arrive at different HRTF contributions and its true hrtf varies significantly by user but the dummy heads objectively attempt to represent a best average. They then ran DBT with trained listeners using similar protocols that were used to arrive at the target speaker response. The limitarions of these tests are the same as for the speaker tests: lack of a true anchor for which to A/B against and the limitations of auditory memory. However in acknowledgement that we don't quite know how room response affects perceived tonal balance, they allowed users to adjust tone controls towards achieving perceived neutrality. They then tested the veracity of their results against external variances in the test subjects, and the outcomes held up well.
None of this is perfect, and the limitations bear repeating: no solid A/B anchors, limitations of auditory memory, assumed IEC room, specific recordings chosen, averaged hrtf. But they tested variance across these variables much more deeply than most studies, and the results held up robustly. Not perfectly, but well. That's an important point to keep in mind. So, eq is always a welcome, necessary tool on the playback end.
You are quite right that individualized HRTF creates the need for eq, but it looks like you're missing the point that the Harman curve gives the best average baseline to eq from, and the best shot at neutral sound for the general population if eq cant be applied.
Individualized Hrtf are great but even if you can test it correctly, you're left with a stark choice:
1. Free field calibration, where the hrtf characterization ignores room contribution. We know thats not quite right
2. Diffuse field calibration that integrates in all the room sound. We also know thats not quite right because a long term average curve ignores how the hearing system uses time windowing to build up a representation of timbre. It will also cause the measurements to inherit the non ideal aspects of your room's acoustics. As Dr Olives headphone study showed, some eq is necessary to tweak the room contribution
I think the Harman outcomes are a major gift to those that seek the most neutral reproduced sound possible. But they do have the limitations discussed. Hopefully future testing will refine these conceptts, but can you propose a better way forward?