Choke works thusly:
Upon bore exit, a shot column will disperse due to aerodynamic effects.
With a cylinder bore (no choke), the drag acts on a wide front of pellets the diameter of the bore. The pictures show radial velocity starting shortly after muzzle exit. The classic 'mushroom'. Dispersion is fairly rapid as the front pellets slow and are pushed to the side by the rear ones.
With a choke, the 'nozzle effect' causes a speed differential, shown as elongation in the pretty pictures. The column is stretched out, and aero forces don't immediately jam pellets to the side as they do with a simple cylinder.
And that's about it.
The WS-1, the Tula, and the Beretta skeet chokes have the interior profile of a converging-diverging nozzle and undoubtedly add an outward radial vector by design.
Inward vector effect is controversial, to say the least.
How did I do, prof?
Better than most for sure. A agree with your first and second paragraphs. Not so sure about the second but I'm not sure what differential" is in reference to - all the shot before it reaches the choke or some of the shot in the choke (the periphery vs the center)? Do you avhe a diagram of the WS-1, Tula, or Beretta skeet chokes?
Seems like folks are arguing for Venturi like effects and claiming a choke works like a de Laval nozzle. I'm not at all sure I buy that, but I see where you are coming from. However, Venturis and de Lavel nozzles are all about the speed of gases or at least fluids. And not about radial dispersion at all. In fact, in the case of the de Lavel nozzle, radial dispersion is to be avoided (so as to generate thrust). But anyway, I'm following along. I just don't see the first principles here working to create the radial spread of the patterns that, undeniably, happen.
Just as a thought experiment considering fluid dynamics (which may or may not be relevant), take your common garden hose. Cut the coupler off the end and run the water on high. What happens? You get a long, relatively solid, stream of water about the size of the inner diameter of the hose for quite a distance before it breaks up. Sort of a full-choke type of pattern, if you will.
Now, put a constriction on the end. A reducing coupling - this acts as a choke. Water comes out faster of course (Venturi), but it also spreads much wider - much more like a cylinder choke in comparison to the first example with no constriction. Clearly, speed isn't everything. In this case, the constriction has exactly the opposite effect as the shotgun choke. But the Venturi principle is preserved, of course, because physics did not go on vacation.
Finally, into that reducing coupling, screw a 28" straight walled tube that is, let's say, 3/4 the diameter of the original hose. Turn the water back on. Now you get a solid stream like the first example, but it is going faster and further. Again, it looks like a full choke but it is, in fact, a cylinder choke, now on a smaller bore barrel.
So these three experiments seem to work exactly opposite of what happens in a shotgun choke. We have all done the hose experiment just fixing up the gardening equipment every spring. And we all have shotguns that we have patterned with different chokes. Clearly, different things happen in each. I am not so sure I trust fluid dynamics to tell me much of anything about chokes in guns. I trust it pretty well in garden hoses.