I am not surprised that they are not well liked. A single esl speaker has such narrow dispersion that all of the reverbs and spacial effects have to come from the recording and not the room.
My assumption would be they are not well-liked when listening to dry monaural material or such with decorrelated reverb, on a single speaker in mono, as they reveal the deficiencies of mono and improper mastering to the fullest. I have done controlled listening tests with different speaker concepts years ago, and electrostatic dipoles were described as revealing the extreme proximity of such recordings.
If you present recordings which contain meaningful spatial cues, like acoustic recordings from a concert hall, this result was flipping, as participants could hear, well, the reverb from the concert hall. With dipoles and their characteristic rear radiation, this even means a certain reduction of perceived proximity compared to a narrow-dispersing speaker w/o dipole pattern.
Are you trying to mount the case that the speaker and room treatment priorities are identical for recording studios and home playback?
Homes and recordings studios are not the same, from the point of an acoustician, and probably never will be. Nevertheless recording and mastering studios are the place where the majority of recordings we listen to at home, are judged and mastered, so keeping our home listening conditions at least roughly close to the average studio one, is a good idea when it comes to decoding the recording as it had been meant to sound.
That said, one of the main differences is that reflections and reverb in studio control rooms tend to be much lower in level and much more balanced in frequency/angle compared to homes, and the variation between different studios is surprisingly small if you compare it with differences between living rooms. So a narrow-dispersion speaker like an electrostat, particularly with even directivity over the frequency range, is one way to come closer to the studio conditions.
The argument is that they have deficiencies that make them less preferred by listeners.
I have organized controlled listening tests involving some stats (no current models in the last 10 years which I found to be improving), and cannot confirm this. Of course there is a pretty vast variation of quality among this speaker category, and obvious limits like dynamic bass capabilities would easily be identified. But if you ask specifically what participants liked in a candidate, almost certainly with an electrostat they would mention transparency, midrange clarity, image stability, non-fatiguing resolution and natural ambience/depth-of-field with acoustic recordings.
The 'science' that I am referring to includes listening tests for preference. But they are controlled.
Preference tests in the meaning of ´do you like the sound of A more than B?´, might be controlled and blind, but the results are from scientific point not worth more than the results of the Pepsi challenge. Well, a majority of participants likes product B over product A under given conditions. That does not allow generalized conclusions on which speaker type or which acoustic property was causing preference, and the result might flip the moment you change one crucial parameter (like channel setup or recording).
When people listen to loudspeakers in the audiophile manner, sighted and uncontrolled, their sound quality impressions are dominated by non-sonic factors.
I have done several controlled listening tests involving stats, and the majority of participants was not audiophile but pro audio. They knew nothing about price, brand reputation and did not care for things like elitism or working principle. They preferred and praised several aspects of the stats´ performance but criticized others, I have no reason to believe that bias was at play (if any, they were skeptical of traditional high end speakers with fancy veneer and piano lacquer).
I was surprised myself how well the electrostats performed overall in these tests, particularly under lively, reverberant conditions. Far better than I would have expected, given technical limitations and poor measurements. Interestingly, the speaker categories that consistently underperformed and met universally harsh criticism despite excellent technical results, were hybrid horns and hybrid electrodynamic/AMT concepts (in which both I had put high hopes).
When told this, ESL owners/defenders start cranking up another list of excuses aimed at discrediting the tests. Which is the typical action of fans of any product when told it doesn't do well in real tests.
Disclaimer: i am neither an ESL owner (have never been) nor defender, but was always reserved and skeptical of stats from theoretical and measurement point. Nevertheless, I had some positive experience with stats and found them surprisingly overperforming in controlled tests.
If you are referring to the Harman institutional tests again: I am not aware of more than one involving electrostats. And it is pretty clear from a neutral position that the particular flaws of this testing method, i.e. monaural ´dry´ recordings presented in a treated room, is a disadvantage for evenly directional speakers, which tend to reveal such properties of the recording (a dry monaural recordings sounds like a ´bird on the wire´ with such speakers which is a good thing from monitoring point).
My understanding is that there has not been any of such tests comparing speaker concepts for almost 20 years, and according to Dr. Olive the whole institution was internally not conducting speaker comparisons for a long time anymore, being eventually discontinued. You can make a wild guess why is that so. Overwhelming success in predicting people's buying preferences under real-market conditions and developing high-end speakers which dominate the market thanks to unbeatable sound quality, does not really count as a reasoning.
I hereby make a wild guess: we are about to occasion some creative excuses by a typical fan of a company that had failed to predict preference under real-world conditions, why the overwhelming majority of an evil consumer base bought products which had been identified as non-preferred in the tests.
