When we are given an unfolded protein to start from, it is given in "extended chain" shape - as close to a straight line as a protein can be. Until I saw the rama map, I believed this to be the proper shape for sheets, so I would always keep sheets in this shape until my design was mostly built. (This may help to explain why my sheets always ended up flatter than I planned.) Also, I would leave any loop segments that I didn't specifically have to change in this shape, and changes such as local wiggles or pulls when closing cutpoints would all start from this shape.
Now that I have seen the rama map, I realize that the extended chain shape is uncommon in nature. I see that using "ideal SS" on a sheet changes its shape and moves all its rama dots to the right and down a little, to a more-populated area of the map. Copying an "ideal loop" in the rama map always involves moving the dots to the right of their starting position. Any time we leave a section of our protein in the given starting shape, we are starting from an unrealistic pose.
I propose that unfolded proteins be given in a more realistic starting shape. One candidate is ideal beta sheet shape (phi, psi = -120, +130). This has the advantage that people who don't use ideal SS on their sheets at the start will nonetheless be starting with good sheets. Another candidate is the other peak in the blue part of the rama map (phi, psi = -65, +145). This represents another linear structure common in nature, a loose spiral called the poly-proline II or polypeptide II (because it doesn't always involve prolines), which is commonly adopted by unfolded proteins. People wanting to make sheets would have to use ideal SS on them to straighten them, which could be seen as either an advantage or a disadvantage. Other moves starting from this shape would start out in a more common area of the rama map and would likely result in more realistic poses.
There is a nice article about the different regions of the rama map at https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3061398/