If you've ever looked at purchasing a pair of loudspeakers you've come across the term "flat frequency response." And in the context of loudspeaker measurement, a flat frequency response at the listening (and also at the measuring) position is the ideal desired result. We all know this. And after a few seconds of looking at a published response curve from a particular loudspeaker (as well as waterfall plots) we have a pretty good idea of how well a loudspeaker can perform in the real world. But with headphones things are not so clearly established...
Before we look at the reason headphone measurement is where it is, let's look at the similarities between headphones and loudspeakers. Both have been around for approximately the same length of time. And both function by moving air in response to an electrical signal. While a loudspeaker works in a room an earphone has your ear as its functional environment. We have developed some extremely sophisticated ways of measuring and then correcting sonic issues in a room, but in someone's ear? Well, that's not as finely-developed a science. Sure, we can measure at the opening of the ear canal, but getting "to the bottom" of the our hearing apparatus isn't a codified process - and even if it was, many humans would object to the invasive nature of such a measurement tool.
So, in short, we have no way to measure and treat ear canals as effectively as we can measure and treat a room. The primary tool for earphone frequency adjustment is by way of digital signal processing where very precise EQ curves can be applied. That's all well and good, but for earphones, measurement is still an issue. Even the most accurate and pricy test rigs can only approximate the perceived frequency response of an earphone above 8 KHz. This is not because measuring devices are insufficiently sensitive, but because variations in humans' ear canals makes their perception of upper frequencies variable from ear canal to ear canal. In short, what you hear from a particular headphone and what I hear could be very different.
For many years the frequency response goals for headphones were quite similar to loudspeakers - a flat frequency response curve over as wide a range as possible. But this began to change. Some of this change was a result of manufacturers and headphone designers seeing that listeners seemed to prefer headphones that had some variations from flat frequency response. Also, every manufacturer has their own proprietary frequency response target curve. Some are based on research, such as the Harman Room-curve, while others are more subjective, but all share the idea that there are better harmonic response target curves than flat.
PSB's Paul Barton, who embraced the research from Canada's National Research Council in his loudspeaker designs, has developed several headphone designs based on the latest work from Sean Olive and Todd Welti that include what PSB refers to as "Room Feel" which adds some additional low frequency equalization that mimics the effects of a loudspeaker in a room. According to research presented by Sean E. Olive, Todd Welti, and Elisabeth McMullin in a May 2013 AES paper, "...the new target based on the in-room loudspeaker response was the most preferred headphone target response curve."
So, according to this research ALL headphones that do not use EQ curves based on their latest research are actually reducing their potential perceived fidelity by trying to be "flat." Does this mean that we should throw out all our old "flat" headphones? Maybe that is somewhat premature.
In my headphone collection I have a number of well-respected headphones that I have used as reference over the years. None have the new EQ curve, but each one does have its own manufacturer's frequency curves, which never measure as flat. And while I have heard both the PSB and NAD headphones that include the Room-feel EQ curve, I did not feel their sonic performance was head and shoulders above competing earphones. Different? Yes. Better? That would depend on your own tastes.
Perhaps someday we will perfect a way to measure and map each individual's ear canal as well as formulate the proper adjustments to compensate for each person's unique hearing environment, but as of today, 2018, we are still making highly educated, but still incomplete, assumptions based on less than ideal measurement techniques and EQ corrections.
Hopefully that will change someday...