My preferred method is to use normal bicubic and then apply whatever sharpening is appropriate after the downsample. I've never liked the fact that you have no control over the amount and type of sharpening in Bicubic Sharper and prefer the two step manual approach. The stairstepped method never really worked better than the one fell swoop. Even in the examples Fred Miranda posted on his website to entice you to buy his actions (as if you couldn't easily write your own) the stepped samples were inferior to a one-step standard bicublc.