more modern magnetic flux designs leading to lower non-linearities, smoother geometries with less edges leading to lower diffraction/directivity issues like the one you mentioned above.
Don´t disagree when it comes down to technical details like magnetic flux, but that does not necessarily translate to different sound quality.
Regarding less edges I tend to look at the whole concept and the acoustic result, not the visual approach of the baffle (which I find quite edgy in KEF Reference and R series btw).
The issue I mentioned was not related to diffraction or baffle geometry, but rather a phenomenon resulting from tweeter diaphragm size and how it is integrated in the midrange voicecoil. It is mildly visible in your graph above 10K and according to my experience negligible if you reduce toe-in of the speaker a little bite. Driver design is a matter of compromise and trade-offs, and I would certainly accept this one rather than broad-banded directivity imbalances.
If you write your preference of such a twice kinked directivity function and a continuously increasing one as wrong it doesn't make it a definition.
I am not aware of any acoustic definition of ´twice kinked´. My aim is to keep directivity as constant as possible in the freq bands which are most localizable by our brain. For the simple reason we can perceive reverb tonality in these bands, and octave-broad (or broader) bands being several Decibels louder or attenuated compared to their neighboring bands, is defined as audible coloration. So even if it might look like a kinked line on the graph, the result is a better tonal balance of the reverb. ´Kinks´ outside this range I consider to be rather unimportant as they can be equalized in most of cases (if there is not ridiculous change in directivity, of course).
their directivities are less different than you think
Quite different in the bands I was referring to, I would say. In particular I mean how the octave-broad bands from like 800Hz upwards compare to their neighboring ones.
On-axis behavior is a different thing. And while I don't deny its importance, I am not paying such attention to it as it anyways have to be corrected in a specific room. An imbalance in the reverb field, after achieving flat on-axis response, is usually not correctible as both are linked to each other in a static way.
