I think one difficulty here is that you need to propose some logical explanation for how "resolution and realism" are based on something other than a speaker's tonality (frequency response), dispersion characteristics/directivity, and distortion performance.
Perhaps dig a little deeper and ask the next question. What exactly does the existing research say about "frequency response?" And that's frequency response at what level of smoothing (1/20th oct? 1/3 oct? no smoothing?)? And under what conditions (anechoic vs your living room, etc.)? How do you weigh the contributions of the different off-axis measurements? What about the slope of the FR plot? Is -1.3 degrees ideal? Can you eyeball a curve yourself, extrapolate the linear regression (in your head), and determine the slope to 1/10-degree accuracy (or even 1 degree accuracy)? You can't (because you're human and not a computer)?
Well, how pray tell, can you eyeball a spinorama and tell me how it will sound (let alone try to do this based on only 1 or 2 charts from a spinorama, like some people are fond of doing)?
As for the "group think here" not fully understanding Harman's research, that might be true of many of us - including you - but Floyd Toole himself has posted here quite recently explaining what the research found and didn't find, and how it did so, and has effectively debunked many of the claims made about the limitations of that research.
Am I a Harman research scholar? Nope, nor do I need to be. But what I have done was read the primary literature (the original papers) myself, thought about them critically, and attempted to correlate the learnings to real life. I put my money where my mouth was and spent over $20,000 purchasing audio products to correlate some of this learning in pursuit of this hobby. And a lot of the Harman research has been absolutely correct. But the way some people interpret it here is just wrong - in research, the term is that it gets "overgeneralized" or misapplied.
People seem to think that research findings are black and white. They usually aren't. Much of the listening preference research has been based on correlations. That means you can say things like "listeners tend to prefer" or "XYZ characteristic explains ___% of the variation in listener preferences." But this research doesn't allow you to say "all listeners should prefer XYZ" or "listeners that don't prefer XYZ speakers are idiots and should have their hearing checked."
That said, I certainly agree with you that different people have different preferences - it's a big world with lots of people so yes, of course. But the Harman research showed that for the most part people's preferences are surprisingly consistent, across geography/culture, age range, and listening expertise or lack thereof.
[/QUOTE]
That's semantics. Harman research actually showed that there is
quite a bit of variability in the quantity of bass and treble preferred across different ages, genders, listening experience, and countries of residence. The official Harman interpretation of that (for obvious reasons, because it translates directly into an engineering goal) is that: "there is a single average preference curve across all humans, we just need to let end users adjust bass and treble quantities to their preference." And that's one valid way of looking at it when you have bass/treble controls. But reading in between the lines, it opens the question of: if there is human variation in bass/treble quantities, certainly there must be human variation in preference of other audible characteristics.
But again, the main issue here is not the Harman research: it's your invocation of the ill-defined notions of resolution and realism.
And you bring up a good point - in that very few measurements (outside of major tilts in the linear regression through an FR curve, which is a no brainer, or "boomy" resonances in the bass region) have been mapped to distinct subjective descriptors of sound and validated through blind listening and publications. It just goes to show the limitations of what we understand when it comes to correlating sound with underlying measurements of transducers.
So, for the purposes of discussion, you can say that my "ill defined notions of resolution and realism" can be translated to "lower listener preference" - i.e. I did not prefer the Genelecs to the HD800 headphones for the two violin recordings. I suppose that if I were to say that "the violins sounded constipated on the Genelecs," that would probably be less accurate.