Ok, I guess I backed myself into a corner, so here goes:
First, there are a lot of opinions, and some have good points, some have bad points, and some are a little of both. Let's start with the reason for a roller rocker to be invented in the first place. It's because they are superior to a stock rocker, period. A mouse trap that does not catch mice is not going to be in existence for very long. Ditto with a roller rocker.
Now, why are they so worthwhile that they are a multi-million dollar industry? It's not because they reduce guide wear from lack of scrubbing action. They WILL reduce scrubbing and guide wear if they are set up properly, but if not, they can scrub just as much as a stocker, and take out the guides just as quickly. Once the roller is under spring pressure and bathed in oil, the roller will slide across the valve tip instead of rolling. A minimum sweep pattern and out-in, out-in motion through the lift cycle will reduce the scrubbing and consequent guide wear.
The real benefit of a roller rocker is the fact that the ratio remains constant throughout the lift cycle. Because the point of rotation is at the centerline of the roller axis, instead of at the contact point on the valve tip, the contact between the roller and the valve tip is always tangent to the axle centerline. That means the ratio is the same everywhere.
Now, we know that the rocker ratio is the length between the shaft centerline and the axle centerline on a roller rocker, divided by the distance between the shaft centerline and the centerline of the pushrod cup or ball in the rocker. That is not the case with a non roller rocker. It is the distance from the shaft centerline, to the CONTACT POINT of the rocker tip on the valve. That is divided by the distance from the shaft centerline to the centerline of the pushrod ball or cup in the rocker.
Look at a non roller rocker as it goes through a lift cycle. The contact point starts inboard on the valve tip, and scrubs across the tip to the outboard side at full lift. Now, where is the contact point on the rocker at full lift? Is it closer to the shaft centerline, further away from the shaft centerline, or the same? Obviously, it is further away, so what effect does that have on ratio? Because the rocker is effectively lengthened, the ratio goes up. That's awesome, right? Well, when you consider that the ratio is set at max lift, it isn't very awesome. It means that the ratio is actually lower everywhere except full lift, which reduces area under the curve, and hurts power versus a properly set up roller rocker of the same ratio.
GEOMETRY! There, I said it. When it comes to off the shelf rockers, there isn't a single one that is designew properly for a Mopar, and it doesn't matter if it comes from someone who specializes in Mopar. The design is a compromise so guys can bolt it on the heads, and stand back and admire how pretty it looks.....until you start it. Proper geometry is dictated by where the shaft is located in relation to the valve tip height. It has nothing to do with how close the roller is to the center of the valve, or how centered the sweep pattern is. It is also determined by the adjuster angle and pivot point height, which is designed into the rocker, and can't easily be changed. Because the shaft location is incorrect for roller rockers on a Mopar head, the rocker adjusters are also incorrect because they are designed for the incorrect shaft position. There is no Harland Sharp this, or Comp that, or Cranes are the hot set up. Not a one is properly designed, and it's unfortunate. The best that can be done, is to relocate the shafts and get some semblance of decent rocker geometry, at least on the valve side.
As far as ratio is concerned, a small block can use a 1.6 in almost every instance. There is a good bit of cam lobe intensity lost through the lifter to pushrod angle and the higher ratio can gain some of that back. For those who have had valvetrain instability after going to a higher ratio, either the geometry wasn't correctly (most likely), or the valve springs were marginal to begin with. A 1.6 on a SBM will almost always result in increased performance.