I don't think this is the right place to have this discussion in depth but there are parts of it that are relevant as Tom's speaker's are all narrower in dispersion.
I only see the comment as stating that different amounts of early reflections result in different perceptions and that those tend to suit certain program material and preferences differently. I don't see how there is a gain from wide dispersion if you then go and absorb the reflections, but I can see that if you wanted to be able to change the presentation through variable room treatment that you would need wide directivity speakers to give you the option.
The definitive quote you posted is not in the 3rd Edition of the book, written as it was in the 1st / 2nd editions. The research is still there and after reading all of it I do not come to the conclusion that wide directivity is always better or more preferred in every situation. What I have found through my own experimentation with directivity and room treatment is that there are indeed trade offs to be had between more or less early reflections and whether they come from lateral directions or not. I find the same as Floyd Toole and Earl Geddes that the program material and personal preference do play a part in which option suits better.
In my view you are taking a very strong stance and using phrases like ignore it or reinterpret it suggests that your view is the only possible way of understanding what was written correctly. For me and many others one size does not fit all. There is no one single solution that works perfectly for all situations, programmes and preferences.
Hi
I would offer that the tall speakers, while a Synergy horn / Paraline lens, the horizontal dispersion is about 180 degrees down to around 500Hz so it would be a very "wide" loudspeaker. So far as room excitation, that is tied to both the H and V angles for the radiation window.
In this case the 20 degree Vertical angle, reduces the room overall excitation.
Also, it's pretty easy to make a speaker that has wide dispersion but it is difficult to make a speaker that has directivity confined to some smaller angle and does that over a wide bandwidth so I am not sure one can make an argument about which does what in a room without having an example of the one you can't make.
Also, making this room sound issue even harder is the fact that loudspeakers usually radiate information from and about the source as well.
By that i mean that if ones eyes are closed, playing a soft voice through just one speaker, that you can easily point the direction of the speaker AND guess how far away the speaker is (playing one speaker).
That depth location identity should NOT happen, the speaker should not add it's spatial thumb print because then in stereo, your aware of the right and left speaker being part of the image with a fixed location as well the phantom image component.
A speaker with simple radiation (like a true point source for example) does not produce spatial clues that make triangulating it's source in depth possible because what reaches your right and left ear is the identical signal.
A source that has close diffraction and other things that produce fine structure comb filtering are common things that radiate this "spatial identity".
Curiously one of the writings of the fellows mentioned it was assured that the interactions between drivers and fine interference wasn't audible. I disagree but one has to hear not read about it.
It is my observation that the less of these extra things speakers do, the more you can make it sound like and even have people get up to look for a center channel speaker to find where the voice in front of them is coming from.
If you can make a 100% solid stereo image anywhere between the two speakers without the speakers identity, there is nothing to stop one from extending that all the way around and overhead with more channels. "All one needs" is a way to capture or create the play back.
That fireworks recording I linked in a post on phase for example is the front two channels of such a system I have been fiddling with.
Best,
Tom