Researcher David Griesinger has studied what makes a great seat in a concert hall. He found that two elements must be in play: First, a clear stream of direct sound; and second, a clear stream of reverberant sound. The key here is that the two be separated in time, rather than smeared together.
Bringing this principle into the home listening room, it can be a challenge to get good differentiation between the direct and reverberant energy streams. The enemy here is, the early reflections. If we can impose a fairly long time delay between the arrival of the direct sound and the onset of significant reverberant energy, we can satisfy the two-streams requirement. This isn't as esoteric as it sounds - Maggie and SoundLab and Martin Logan and Quad owners do it all the time! By pulling their fairly directional dipole speakers out into the room a good five feet or so, they are imposing roughly 10 milliseconds of delay on the backwave energy, because of the path length of its bounce off the wall. The 10 millisecond target comes from my own observations as a longtime dipole user, and is supported by Earl Geddes, who is a leading expert in the field of small-room acoustics and psychoacoustics.
(A quick tangent is in order: In home audio systems, there is often a fundamental problem that needs to be addressed before we start worrying about meeting the requirements of the two-stream paradigm, and that is, the spectral balance of the reverberant field is usually wrong. The vast majority of energy in the reverberant field starts out as off-axis sound, so we need to get the off-axis response right. In general most omnis, dipoles, and other polydirectional speakers do a better in this area than most conventional cone-n-dome speakers. Many speakers with rather aggressive pattern control (good horn or waveguide based systems) also do well in this area.)
That 10-millisecond delay target is harder to meet with wide-pattern cone-n-dome speakers or omnis because they have more near-side-wall interaction. And most speakers will have some floor and ceiling bounce, which fortunately the ear is not especially sensitive to despite the fact that they happen well before our 10 millisecond target.
Personally I am not in favor of the overly aggressive use of absorption to soak up the energy in the early reflections. Absorption is always more effective on short wavelengths than on longer ones, and this absorption happens to all reflections that hit the absorptive material, not just the first ones. So the net effect is to roll off the high frequency energy in the reverberant sound, which is usually lacking in high frequency energy as it is.
In my opinion the spectrally correct (and obviously well-energized) reverberant field of a good omni is beneficial, and especially so for classical music because it more closely approximates the concert hall experience. But if we want to get closer to the experience we'd have in a really good seat in a concert hall, we should look at the two-stream paradigm, and try to get a clear time-differentiation between our direct and reverberant sound streams. Imo dipoles, bipoles, and other well thought-out polydirectionals can deliver the goods if they are set up correctly. And if we have the two-stream paradigm (and that 10-millisecond target) in mind, imo we are on the right track.