Rail compression is not required to explain the shot.
It is enough to have friction between the ball and the cushion as the ball moves
somewhat sideways across the cushion.
Can you think of an experiment that would distinguish between rail compression
(which seems to be something like the bunching up of the cushion material forward of the cue ball)
and simple rail-ball friction?
Well, I agree that rail-induced sidespin helps the ball reverse direction. That explains its direction off the 2nd rail.
I just think wanted to point out that rail compression is necessary to shorten up the ball enough,
for this induced sidespin to be be useful in getting the ball 'backwards'... backwards enough to go into a pocket.
Without the shortening caused by rail compression, the induced sidespin would try to bring the ball back,
but I don't think it could ever be enough to pocket the ball.
This bank would illustrate it better I think -
https://www.youtube.com/watch?v=ytPZcWNa2Ao
With lines shown:
https://i.imgur.com/u1OFBR2.jpg
If the ball didn't shorten a ton off the first rail (near the meucci sticker) I think it hits the 2nd rail around the end of the yellow line.
We know that the induced spin changes its angle off the 2nd rail. By how much? By about the difference between the light green line,
and the dark green line. That's a healthy change in rebound angle, but if the object ball reversed by that amount from the end of the yellow line...
it could never go backwards enough to get near either pocket. So it's a combination of shortening and rail-induced spin. And of the 2 I think the shortening plays a bigger role.
My other thought, which is just a guess, is that the "cupping" of the ball by the rail gives it more surface area in contact with the ball, which in turn creates more grip to turn the ball. Like so:
As for experiments, I am having trouble thinking of one that removed rail compression from the equation.
I guess wooden rails, or putting pool cloth on a floor and wall and observing how much spin transfers when it is rolled at an angle into the wall.
Not 100% relevant but do you recall ever seeing this demonstration of collision-induced sidespin? I remember being taught this as a kid.
I cannot recall the exact CB starting position. But the idea is clip the 1 as thinly as humanly possible.
The CB picks up sidespin and reverses off the 2nd rail, with minimal force.
https://pad.chalkysticks.com/5589d.png