Indulge me while I theorize for a moment.
A critical moment in the travel of a patched round ball is when it makes the transition between barrel and free air. One microsecond it is compressed in a patch and guided by the barrel - the next it is in free air with the patch falling away behind it. Obvious enough.
But what happens at that transition, when the band of contact around the equator of that ball is emerging from the bore? In a perfect world the ball, patch, and barrel are all absolutely symmetrical, and a flawless ring-shaped blast of hot gas emerges like an expanding doughnut behind the ball.
In an imperfect world, the patch is a bit uneven in thickness, weave, and lubrication. The ball is slightly out of round, as is the bore. That ring of gas behind the ball will be lumpy. The gas will start escaping sooner on one side than the other, or perhaps at several places. The ball will acquire a slight rotation or skew.
The more symmetrical the elements and the shorter that period of transition time, the smaller this effect will be. That's why people measure their patching material, use swaged balls, and use false muzzles to prevent wear. The perfect muzzle would be a sharp 90 degree cut, but try to load into that without a false muzzle.
Coning spreads out that transition, giving more time for any asymmetries to affect the flight of the ball. In a perfect world there would be no difference between a coned muzzle and a 90 degree muzzle, but that is not where we live. Of course, it is a continuum between a crown and a cone. How deep does a crown have to be before you call it a cone?
A shooter with consistently accurate components and a well coned barrel may never notice the effects of coning. A bench rest shooter undoubtedly would, because the margin of error is smaller in that competition.
It comes down to 1) the precision and symmetry of the coning, and 2) the consistency of the ball and patch. A coned bore would be more sensitive to variations.
Personally, I am going to stick with a crown and allow myself greater variability of components.
Canute