Paul,
I went back to re-read my article and looked at all the photos I took during the process so I was sure to give you accurate information.
In a nutshell, I simply wiped off the old finish to the point where I wanted to re-start, you don't need to go to bare wood. This process, if done as described in the article, leaves you in complete control. I must stress the part about working in small areas, and not let the thinner run all over the model. Have a goal for a session, like the underside of a wing panel, then stop to re-group.
On the wings, I decided that pulling off the silkspan completely would make the edge of the fillets and hinge lines an issue for refinishing, so I left the covering on all the sheeted/solid portion of the wings and stripped the paint down to the silkspan over all balsa surfaces (including capstrips). I never pulled off any silkspan. I left the open bays alone and carefully cut those sections out later, leaving the capstrips, TE, LE, and Center sheeting covered. The edges of the capstrips and sheeting were then carefully sanded with a small radius.
I re-covered the open bays of the wings and tore the edges of the silkspan against a metal straight edge so that it would not have a scissor cut hard edge and would then blend easier with the sheeted center sections and LE sheeting that were already covered (only over lapping a 3/4 inch or so).
On this model, the flaps and tail surfaces were covered in carbon veil, and those surfaces did not fair as well to the stripping process so they were re-covered with light silkspan over what veil remained.
The Fuselage needed the most modification. I re-shapped the cowl, replaced the canopy, reshaped vertical stabilizer as well as belly of the fuse. In spite of all that, the tissue was preserved around the fillets and new silkspan was applied everywhere with soft, torn edges to make blending easier. I never pulled tissue off a covered surface, rather I blended them with new tissue.
I hope this helps, I would not hesitate to do it again. PM me if you would like to talk about the process over the phone. To be clear, this is not my method. Credit and thanks must go to Randy Smith and Frank McMillan for walking me though the process.
Curt