I know this is old stuff, and I didn't read every page so pardon me if I'm reposting known info, but this article in Circle Track is at variance with the well-known Comp Cams page.
Rocker Arm & Valve Train Geometry - Circle Track Magazine
It references both Miller and Smokey Yunick, and has some great drawings which show why a narrow roller track means more valve lift than a wide track.
I also emailed Comp on their diagram last year sometime, and never got a reply.
All this having been said, I can't imagine anybody who's, e.g., changing rocker arms, pulling the heads to change valve height (guess lash caps might work if you're lucky) just to get an extra .005 lift on a street motor. But the philosophy of a short roller track across the valve tip is correct, if real-world balanced against not having the pattern too far off-center. Note that the shorter the track, the closer you get to the ideal of having the exhaust-side limit of the roller track occur at mid-lift (not at full lift as Comp shows). Dead-centering really isn't necessary, nor IMO is having the lifter fully primed for checking unless you're racing and want every last .001 at any cost. Else, it's perfection versus the law of diminishing returns here.