How about breaking it down to the theoretical fundamentals: What would be best if it could be perfectly realized in physical implementation (an impossibility)?
>A completely omnidirectional, evenly radiating pulsating sphere (or point, if you prefer)
>Same as the above, but a monopolar hemisphere only, maybe wall-mounted to simulate an infinite baffle constituting a listening room boundary
>A true monopolar, laser-like (parallel-focused) 'ray of sound' eminating from a point-source (actual or simulated) and aimed directly at one of the listener's ears (what would the difference be between this and a set of headphones?)
>A 360-degree radiating, cylindrical line-source with zero vertical dispersion
>Same as above, but a 180-degree cylindrical half-section only ('monopolar' line-source, analogous to the sphere/hemisphere example above), again maybe wall-mounted to simulate an infinite-baffle
>A dipolar version of of any of the above, such as a Quad ESL is to a simulated pulsating sphere
>None of the above: The ideal radiation pattern should be an exact inverse of the recording microphone's 'acceptance-field' pattern, whatever that may be (in realistic terms, this kind of thinking could only even begin to apply with a very tiny minority of recordings actually made, due to prevalent recording methods)
>None of the above: Given the preceeding, stipulate that you'll never be able to standardize and optimize the recording process to conform to some idealized encode/decode protocol with an equally-conforming playback system, and thus there is no one 'correct' ideal radition pattern possible, so you should just work with whatever sounds good to you in your room
>It doesn't matter, as long as you listen inside a perfect anechoic chamber, maybe in a multi-channel setup
>A completely omnidirectional, evenly radiating pulsating sphere (or point, if you prefer)
>Same as the above, but a monopolar hemisphere only, maybe wall-mounted to simulate an infinite baffle constituting a listening room boundary
>A true monopolar, laser-like (parallel-focused) 'ray of sound' eminating from a point-source (actual or simulated) and aimed directly at one of the listener's ears (what would the difference be between this and a set of headphones?)
>A 360-degree radiating, cylindrical line-source with zero vertical dispersion
>Same as above, but a 180-degree cylindrical half-section only ('monopolar' line-source, analogous to the sphere/hemisphere example above), again maybe wall-mounted to simulate an infinite-baffle
>A dipolar version of of any of the above, such as a Quad ESL is to a simulated pulsating sphere
>None of the above: The ideal radiation pattern should be an exact inverse of the recording microphone's 'acceptance-field' pattern, whatever that may be (in realistic terms, this kind of thinking could only even begin to apply with a very tiny minority of recordings actually made, due to prevalent recording methods)
>None of the above: Given the preceeding, stipulate that you'll never be able to standardize and optimize the recording process to conform to some idealized encode/decode protocol with an equally-conforming playback system, and thus there is no one 'correct' ideal radition pattern possible, so you should just work with whatever sounds good to you in your room
>It doesn't matter, as long as you listen inside a perfect anechoic chamber, maybe in a multi-channel setup