Ok, lots of confusion here.
I did some experiments awhile back comparing sightline and bore line for revolvers vs. Browning type semi-auto pistols and found that they are quite different.
Revolvers are set up to compensate for significant muzzle rise between the time that the trigger is pulled and the bullet exits the bore. The same would be true of any fully locked breech action like a gas operated action (such as the Desert Eagle) or any manual repeater as long as the boreline is above the point of resistance caused by the shooter's grip.
But linkless Browning type semi-auto pistols are set up differently. The borelines and the sightlines are almost perfectly aligned. In other words, they are not set up to compensate for any significant muzzle rise between the time that the trigger is pulled and the bullet exits the bore.
Recoil DEFINITELY begins when the bullet starts moving. There is no room for debate on that point.
It's important to understand that muzzle rise and recoil are two different things.
Recoil is an unavoidable consequence of bullet movement.
Muzzle rise
sometimes results from recoil
depending on the circumstances. And it may not happen immediately.
Recoil does not always cause muzzle rise. If the recoil vector is straight back into the point of resistance, the recoil motion will be straight back. If the recoil vector is above the point of the resistance, the recoil will cause muzzle rise. That's the first complication.
In addition, even when recoil does cause muzzle rise, it does not always have to cause
significant muzzle rise
immediately. In the case of the linkless Browning type semi-auto pistols, the recoil initially causes the slide and barrel to move backwards, resisted only by the force of the recoil spring. That remains true, by design, until the bullet exits the bore and the barrel and slide unlock. Because the recoil spring presents a relatively small resisting force, the muzzle rise during this phase of recoil is minimal--the slide and barrel move pretty much straight back with no significant force being transferred to the frame of the gun. We can see that this is true by looking at the sightline vs. boreline for this type of pistol and noting that there is no attempt to compensate for muzzle rise.
The muzzle does rise a very little bit due to the loose coupling of the slide/barrel to the frame via the recoil spring, but not very much at all--certainly not nearly as much as would be seen in a revolver, or in a gas operated pistol like the DE.
This post contains a very thorough analysis of just how much the muzzle of a Browning type action might rise while the bullet is in the bore.
Once the slide and barrel travel backwards far enough to reach the unlocking point, the barrel stops against the frame (or against something that bears against the frame) and that provides the first direct coupling of the recoil energy to the frame of the gun. At that point muzzle rise begins to really show up. However, it's important to remember, that, by design, the linkless Browning type semi-autos do not unlock until the bullet has left the bore. So the really significant part of muzzle rise happens after the bullet is gone.
Browning type pistols with links that begin tilting the barrel immediately when recoil starts are sort of a different category and I haven't done measurements on them.