DF merely is a measurement method where a HATS is used and the final (compensated) result is created from multiple speakers in an anechoic room.
By lack of other measurements (FF is even more wrong) headphone manufacturers chose the DF to be closest to a headphone.
It isn't but it is better than FF.
Harman took their ref speakers in their 'optimal listening room' and their HATS with their pinna. Measured the FR in that room and the familiar Harman curve we see so often thus is an 'averaged' (weird dips and peaks 'smoothed') is the result.
When you get headphones to follow that 'average' (you would have to smooth it to the curve which no one does) then you are close to speakers measuring flat in a room. For headphones some extra lows is added to make up for the lack of tactile feel.
Seems like a better standard than something designed for 'measuring' omni directional SPL using a HATS.
Yes, in theory the Harman target makes more sense, and it is difficult to argue with the science. However, I have a few problems with it:
1. It is based on people's preferences, and in that sense is a poor indicator of realism. I am sure the vast amount of folk who listen to crap music that has been poorly mastered probably prefer the Harman curve, because it is more forgiving of bad material, it masks imperfections, and it results in a more pleasing/dramatic sound in those cases.
2. I have empirical evidence that DF tuned headphones sound much more realistic than Harman tuned headphones. The detail, transients, soundstage, everything is superior. Percusive instruments, including cymbals, are fully represented, all the harmonics in stringed instruments are there (as far as my hearing allows me to distinguish).
I can't bring myself to argue against point #2, and I don't care much about the science behind it, to be honest. I only care about the result. I understand why for a company like Harman, and most other manufacturers, using Harman makes sense, since that appears to be what the majority likes. But I am not the majority, and I don't care what the majority likes. I am not contesting the science and the throught behind this, but the result is all that matters to me.