Because clearly you've never installed a system that achieves cross-talk cancellation. Did you see the algorithm I stated? Have you ever installed a system that produces those signals at the left and right speakers with them spaced at the angle stated, in conjunction with a L+R center speaker? If not, then how can you say it's not possible?
As far as dispersion angle, I've used only 'normal' speakers with typical dispersion patterns; the cross-talk cancellation is achieved by a combination of time delay and opposing-channel signal inversion, not controlled dispersion patterns.
Just because you've never heard cross-talk cancellation doesn't mean it doesn't exist.
https://phaidra.kug.ac.at/open/o:11444
I've installed three such systems in quite different rooms and they all produced nearly complete cancellation of crosstalk. The sensation is actually quite uncomfortable with a signal present at only the left amplifier channel (for instance), the sensation is of having gone suddenly deaf in the right ear; the right ear hears nothing of what's coming out of the left speaker because the crosstalk is completely out of phase with the direct sound and so nulls it at the right ear. The wide spacing is just sufficient to produce the required time delay for this null to occur; it's not just putting speakers
really far apart, in fact anything beyond 120 degrees apart doesn't result in the correct delay for the cancellation.
The three different versions of this I've done are:
1. Simple passive with wide range single-driver speakers (think Auratone, with much higher quality drivers).
2. Bi-amped 4" two-ways, with the tweeters passive-connected to one stereo amp, and the bass-mids to a second amp.
3. Multi-channel version (which can be used with passive-crossover speakers) utilizing a very simple analog matrix circuit consisting of basic inverters and summers, that produces the required separate signals for each amp channel.