At a certain point any graph, measuring result or scientifically accurate observation will fail to describe the human experience, especially one that corresponds to and is so fundamentally influenced by something as subjective as emotion.
I totally agree that measurements aren't predictive or even correlated with my emotions when I listen to music. I can verify this every time I listen to music. I am lucky enough to have 4 decent systems in 3 different rooms and I can't even predict very accurately which one will work for me at any given time. Sometimes, I sit on the couch, fire up system 1 and things sound wonderful. Sometimes I listen to system 1 for 5 minutes, tell myself it isn't working as well as I hoped, and switch to system 2. Sometimes I switch rooms. On random days, everything sounds wonderful, or unsatisfying. l may one day miss the deep bass my subs provide or find it too much a bit later. About the only time there is a correlation between measurements and my emotions is when I spend some time measuring stuff and get the result I am aiming for. I feel happy and satisfied if I reach my goal, frustrated if I don't. So, yeah, one can say current audio science is limited on that front. It could be that all science needs to predict the outcome of my listening session is a complete panel of all my neuro-transmitters at a given moment and an fMRI, but that is hardly practical.
Subjectivity does indeed rule the audio experience.
However, the problem with that fact is that it is useless. My subjective experience is not your subjective experience. If I told you my system 1 had such "liquid mids it brought tears to my eyes", what would you make of it in practice? Replicate my system 1? Replicate my room and furniture? My ears? My instantaneous mindset?
That being said, all measuring hope is not lost. One can, for example, test many speakers with many listeners and come up with a result that shows what most people seem to prefer. One can analyze the results and identify the factors that seem to influence the preferences. It doesn't tell us what you or I should prefer or what you and I actually prefer. At best, it gives you a probability of what you might prefer, all other things being equal. That's a tiny piece of information, not a rock-solid predictor, but it is better than what a random subjective reviewer can provide, even an honest subjective reviewer (it is hard to think of one with a significant audience that doesn't do product placement).
As a listener, your subjective experience is 100% of what matters. You don't care about what the science says, unless knowing your measured system performance is, by itself, a source of emotional satisfaction. But the fundamental mistake even honest to God subjective reviewers make is to believe that their subjective experience is valuable to others, worth fighting for. Your subjective experience is yours and yours alone.