That is weird now that you point it out. Do you think that would change with a high end Revel or JBL M2? Seems like Genelec beats them at their own game too, and after listening to the LSR series, and the 705, I wonder if $20,000 is enough to get good JBL speakers or if it's just a louder, (they would probably call it 'more dynamic') version of their 7 series without much bass extension relative to the size, good horizontal dispersion like everyone else, poor vertical like everyone else, and archaic cabinet design like everyone else, but with additional braces because they can afford more glue and wood with the price tag.
If Buchart and Genelec had the market and economies of scale JBL had, I imagine they'd be able to get prices even lower and further compete wkth Harman on the low end and high end.
I mean, it's pretty simple really: My guess is Harman doesn't care about the preference score all that much. I would not be surprised if they don't even regularly calculate the score when designing speakers. The score is only an approximation for blind tests, which they
do conduct, presumably more than anyone else.
There are principles from that score research theu take into account. The best speakers will have flat direct sound and smooth directivity -- not nessarily in the DI curves, but naturally a speaker with flat on axis and smooth PIR will score well.
But from Amir's measurements and others, it's become evident to me that Harman tends to prioritize smooth and wide horizontal directivity consistently for their home speakers even if sometimes at the expense of the vertical which shows up in the PIR as a crossover dip.
Look at the HDI-1600: it has a deep waveguide, which would normally mean narrow horizontal directivity, yet JBl seems to have gone out of it's way to give it wider horizontal directivity than you'd expect.
Same with the revel line: the Salon2 is not the flattest or smoothest speaker even among other revels, let alone some other brands. But what it does really well is very wide horizontal directivity courtesy of the shallow waveguide and sculpted baffle.
There are probably other things at play, like optimizations for specific frequencies, but this is the main quality I've noticed. It's hard to beat a good coaxial when it comes to vertical reflections and therefore smooth spinoramas, but clearly Harman as a group doesn't think the benefits outweigh the typical SPL and directivity limitations.