There is an uncertainty in the readings from the instruments regardless of the technique or which instrument we are discussing.
What gets frustrating is learning to accept a non-zero uncertainty, and learning when uncertainty is too high due to other imperfections in the ammo.
Part of the uncertainty is caused by the samples we are looking at and the fact that some have low runout on the base of the case and also on the runout of the bullet, and some don't.
The nature of trying to settle the gage pressure on a long cartridge using a narrow base against the point where the intersection of an ogive and a circle are minimized, depends on cartridge runouts and the gage pressure.
That straight line may or may not be the stiffest if the base doesn't represent the centerline of the cartridge. If it is tilted at all, the tip of the rim is loaded and it isn't as stiff as one where the base area is fully supported.
If a tool is very square to a perfect bullet, then the intersection is a circular ring. But, what happens when a bullet is not straight?
That intersection starts to be more like an ellipse, but the line of action of the tool is only getting supported on the long axis of the ellipse.
So for a given gage pressure, the measurement uncertainty will grow when there is misalignment in the base or bullet.
If the ammo is sorted by runout, including the base of the cartridge and the bullet centerline, you will notice a more repeatable reading with lower uncertainty.
The opposite is also true, try the same readings on samples with bad runouts and the CBTO readings get more inconsistent too.
If you are getting an uncertainty of +/-0.001, you are doing pretty good.