OB sets out to address a very specific problem: uncontrolled directivity in conventional speakers. Briefly, this refers to a typical box speaker radiating sound in all directions at low frequencies and only radiating sound forward at high frequencies. You hear more than just the direct, forward radiated sound. You hear the contribution from room reflections as well. If those reflections from side walls, front wall etc do not have similar frequency response to the direct forward radiated sounds, our brains recognize the mismatch and this can be interpreted as skewing the perceived frequency response, or worse a contradiction that diminishes the realism of reproduction. Another way to state this problem is that the power response is not even, ie the energy radiated at all angle is not smooth and flattish wrt frequency.
You can attack this problem in three ways... correct the way the speaker radiates sound, alter the way your room reflects sound, and hold your head in a vice while nonlinear corrections are applied to the signal before amplification by a dsp to "undo" the speaker and room deficiencies. While the latter two can be done in theory, the first is preferred with results being applicable go a larger seating area, the room next in preference, and the digital correction fun but IMO best suited for single listeners and/or modest manipulations.
So a dipole is one way to tackle speaker directivity. The out of phase sound front and rear cancels at the sides, leaving a null. In ideal circumstance, this cancellation is uniform regardless of frequency and the directivity is perfectly constant with frequency having a somewhat narrow figure of eight pattern. Real world is not ideal, drivers have real and varying size and geometry, and the cancellation is not perfectly uniform. But its pretty danced good, the problems mostly happening in the upper frequencies (several kHz) where program content is dropping and these cues are less important psychoacoustically.
One obvious tradeoff is efficiency. The cancellation lowers output in a frequency dependent way, so low frequencies drop like a rock. Building a bass dipole is difficult... a true dipole sub typically impractical. To achieve reference levels to workable frequency, say down to 80hz or so where integration with monopole sub is eased by change in room behavior, will probably require a three or four way with multiple drivers covering lower midrange and bass frequencies. So you get a large, expensive speaker.
You would think the backwave is a nuisance, but it isn't. Even with the backwave, a dipole emits less total sound into the room for a given on axis spl than a comparable monopole box speaker, so thee is less contribution to the sound from room reflections, not more. If you want to absorb the backwave high frequencies that isn't difficult, leaving you still with less rear radiated low frequency information than a monopole but a better controlled front radiation. Still a win. Some also cite the reduced sound that is reflected from a box internally and is radiated back through the speaker cone as a source of improved sound quality. IMO this is a minor factor.
Since there is less sound radiated into and reflected from the room for a given on axis spl, room treatment requirements may be reduced. This might be particularly helpful in a casual environment with lots of reflective surfaces. On the other hand, there is research showing many people like side wall reflections, and dipoles produce less of that, so this could be a negative rather than a positive depending on your personal preference. That research also says people prefer reflections with frequency response similar to the on axis response rather than far different as is common with many box speakers, which dipoles certainly improve greatly on. These are conflicting forces that are preference driven. In a multichannel system, less reliant on and influenced by lateral reflections, IMO he potential downside of narrow dispersion is reduced.
Speaking of narrow dispersion, some also cite that as negative if you want to have multiple seating positions. Perhaps. It is often possible however to toe in dipoles so that the on axis lines cross in front of the main listening position, which then becomes "off axis" to both left and right speakers. Interestingly, now the left of center seats are further from the right speaker but on axis, and closer to the left speaker but more off axis, which due to the narrow dispersion can compensate some of the volume imbalance from being nonequal distances to left and right and reduce skew of imaging. This is the same technique advocated by geddes for his speakers which have a similarly narrowed dispersion. Note that this is only possible if dispersion is narrow, and only advised if off axis response is similar to the on axis response since the primary listening position is off axis somewhat to both left and right. That is common for good dipoles, uncommon even for good monopoles.
The primary alternatives for controlling speaker power response are omnipoles and horns. Omnipoles have nearly opposed tradeoffs compared to dipoles... they have the lowest direct to room reflected sound ratio, are easy to make low frequency omnipoles but difficult to make good mid and high omnipoles, and have the greatest interaction with the room. Horns can do lots of things right, offering good uniform front dispersion with no backwave, have somewhat wider front dispersion, and are very efficient which reduces some forms of distortion. They have their tradeoffs like any other design, primarily being similarly difficult to extend down low like dipoles, the physical size of the horn necessitating large driver to driver center distances causing issues with driver integration in domestic sized rooms, and the tendency for the horn to be more efficient at amplifying overtones than the fundamental in many cases.
But this isn't the thread for omnipoles or horns or finite impulse response room corrections.